{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 13873,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 7.208246233691343e-05,
      "grad_norm": 10.785984856921953,
      "learning_rate": 4.796163069544364e-07,
      "loss": 5.7781,
      "step": 1
    },
    {
      "epoch": 0.00014416492467382686,
      "grad_norm": 8.315240311974708,
      "learning_rate": 9.592326139088728e-07,
      "loss": 5.3225,
      "step": 2
    },
    {
      "epoch": 0.0002162473870107403,
      "grad_norm": 8.669449923784002,
      "learning_rate": 1.4388489208633094e-06,
      "loss": 5.8127,
      "step": 3
    },
    {
      "epoch": 0.0002883298493476537,
      "grad_norm": 8.471233834791175,
      "learning_rate": 1.9184652278177456e-06,
      "loss": 5.758,
      "step": 4
    },
    {
      "epoch": 0.00036041231168456716,
      "grad_norm": 8.28458835686889,
      "learning_rate": 2.3980815347721824e-06,
      "loss": 5.6031,
      "step": 5
    },
    {
      "epoch": 0.0004324947740214806,
      "grad_norm": 8.941649352104648,
      "learning_rate": 2.877697841726619e-06,
      "loss": 5.4906,
      "step": 6
    },
    {
      "epoch": 0.000504577236358394,
      "grad_norm": 10.922813857662273,
      "learning_rate": 3.3573141486810552e-06,
      "loss": 5.3222,
      "step": 7
    },
    {
      "epoch": 0.0005766596986953074,
      "grad_norm": 7.48706450481227,
      "learning_rate": 3.836930455635491e-06,
      "loss": 5.5716,
      "step": 8
    },
    {
      "epoch": 0.0006487421610322209,
      "grad_norm": 10.012903824976998,
      "learning_rate": 4.316546762589929e-06,
      "loss": 5.3152,
      "step": 9
    },
    {
      "epoch": 0.0007208246233691343,
      "grad_norm": 10.125081420118354,
      "learning_rate": 4.796163069544365e-06,
      "loss": 5.8179,
      "step": 10
    },
    {
      "epoch": 0.0007929070857060478,
      "grad_norm": 8.736966060522292,
      "learning_rate": 5.275779376498801e-06,
      "loss": 5.3012,
      "step": 11
    },
    {
      "epoch": 0.0008649895480429612,
      "grad_norm": 7.307969573939186,
      "learning_rate": 5.755395683453238e-06,
      "loss": 5.7993,
      "step": 12
    },
    {
      "epoch": 0.0009370720103798745,
      "grad_norm": 7.376442120543025,
      "learning_rate": 6.2350119904076745e-06,
      "loss": 5.794,
      "step": 13
    },
    {
      "epoch": 0.001009154472716788,
      "grad_norm": 8.320001382592888,
      "learning_rate": 6.7146282973621105e-06,
      "loss": 5.7717,
      "step": 14
    },
    {
      "epoch": 0.0010812369350537015,
      "grad_norm": 8.50741635840794,
      "learning_rate": 7.1942446043165465e-06,
      "loss": 5.174,
      "step": 15
    },
    {
      "epoch": 0.0011533193973906149,
      "grad_norm": 7.287595923403942,
      "learning_rate": 7.673860911270982e-06,
      "loss": 5.1793,
      "step": 16
    },
    {
      "epoch": 0.0012254018597275282,
      "grad_norm": 6.9368526606823115,
      "learning_rate": 8.153477218225421e-06,
      "loss": 5.5536,
      "step": 17
    },
    {
      "epoch": 0.0012974843220644417,
      "grad_norm": 7.739065819867966,
      "learning_rate": 8.633093525179858e-06,
      "loss": 5.1211,
      "step": 18
    },
    {
      "epoch": 0.001369566784401355,
      "grad_norm": 6.505267141975724,
      "learning_rate": 9.112709832134293e-06,
      "loss": 4.9048,
      "step": 19
    },
    {
      "epoch": 0.0014416492467382686,
      "grad_norm": 7.440258242103564,
      "learning_rate": 9.59232613908873e-06,
      "loss": 4.929,
      "step": 20
    },
    {
      "epoch": 0.001513731709075182,
      "grad_norm": 7.187601732367615,
      "learning_rate": 1.0071942446043167e-05,
      "loss": 5.2239,
      "step": 21
    },
    {
      "epoch": 0.0015858141714120955,
      "grad_norm": 6.77828456789048,
      "learning_rate": 1.0551558752997602e-05,
      "loss": 5.2967,
      "step": 22
    },
    {
      "epoch": 0.0016578966337490088,
      "grad_norm": 6.896217729362742,
      "learning_rate": 1.1031175059952039e-05,
      "loss": 4.8037,
      "step": 23
    },
    {
      "epoch": 0.0017299790960859224,
      "grad_norm": 7.008963177389828,
      "learning_rate": 1.1510791366906475e-05,
      "loss": 4.8777,
      "step": 24
    },
    {
      "epoch": 0.0018020615584228357,
      "grad_norm": 7.984975718727849,
      "learning_rate": 1.1990407673860912e-05,
      "loss": 4.8595,
      "step": 25
    },
    {
      "epoch": 0.001874144020759749,
      "grad_norm": 5.723942495634698,
      "learning_rate": 1.2470023980815349e-05,
      "loss": 4.2427,
      "step": 26
    },
    {
      "epoch": 0.0019462264830966626,
      "grad_norm": 6.280311969269948,
      "learning_rate": 1.2949640287769784e-05,
      "loss": 4.3036,
      "step": 27
    },
    {
      "epoch": 0.002018308945433576,
      "grad_norm": 6.24523416099905,
      "learning_rate": 1.3429256594724221e-05,
      "loss": 4.3513,
      "step": 28
    },
    {
      "epoch": 0.0020903914077704893,
      "grad_norm": 6.266680853288844,
      "learning_rate": 1.3908872901678658e-05,
      "loss": 4.3505,
      "step": 29
    },
    {
      "epoch": 0.002162473870107403,
      "grad_norm": 6.462571473872499,
      "learning_rate": 1.4388489208633093e-05,
      "loss": 4.059,
      "step": 30
    },
    {
      "epoch": 0.0022345563324443164,
      "grad_norm": 7.678373277722772,
      "learning_rate": 1.486810551558753e-05,
      "loss": 4.2533,
      "step": 31
    },
    {
      "epoch": 0.0023066387947812297,
      "grad_norm": 5.5230275766356565,
      "learning_rate": 1.5347721822541965e-05,
      "loss": 3.8493,
      "step": 32
    },
    {
      "epoch": 0.002378721257118143,
      "grad_norm": 5.834271002297824,
      "learning_rate": 1.5827338129496403e-05,
      "loss": 3.9212,
      "step": 33
    },
    {
      "epoch": 0.0024508037194550564,
      "grad_norm": 5.803278608983696,
      "learning_rate": 1.6306954436450842e-05,
      "loss": 3.753,
      "step": 34
    },
    {
      "epoch": 0.00252288618179197,
      "grad_norm": 5.605833795522804,
      "learning_rate": 1.6786570743405277e-05,
      "loss": 3.7069,
      "step": 35
    },
    {
      "epoch": 0.0025949686441288835,
      "grad_norm": 4.8902826506136305,
      "learning_rate": 1.7266187050359716e-05,
      "loss": 4.2436,
      "step": 36
    },
    {
      "epoch": 0.002667051106465797,
      "grad_norm": 6.296770871639688,
      "learning_rate": 1.774580335731415e-05,
      "loss": 4.0908,
      "step": 37
    },
    {
      "epoch": 0.00273913356880271,
      "grad_norm": 5.22757354730745,
      "learning_rate": 1.8225419664268586e-05,
      "loss": 3.6499,
      "step": 38
    },
    {
      "epoch": 0.002811216031139624,
      "grad_norm": 4.567886902486846,
      "learning_rate": 1.8705035971223024e-05,
      "loss": 3.6501,
      "step": 39
    },
    {
      "epoch": 0.0028832984934765373,
      "grad_norm": 6.853341525149152,
      "learning_rate": 1.918465227817746e-05,
      "loss": 3.4999,
      "step": 40
    },
    {
      "epoch": 0.0029553809558134506,
      "grad_norm": 5.9467064118574475,
      "learning_rate": 1.9664268585131895e-05,
      "loss": 3.2221,
      "step": 41
    },
    {
      "epoch": 0.003027463418150364,
      "grad_norm": 6.4980768175094825,
      "learning_rate": 2.0143884892086333e-05,
      "loss": 3.3054,
      "step": 42
    },
    {
      "epoch": 0.0030995458804872773,
      "grad_norm": 4.326060557387276,
      "learning_rate": 2.062350119904077e-05,
      "loss": 3.1091,
      "step": 43
    },
    {
      "epoch": 0.003171628342824191,
      "grad_norm": 3.057761939416834,
      "learning_rate": 2.1103117505995203e-05,
      "loss": 3.3289,
      "step": 44
    },
    {
      "epoch": 0.0032437108051611044,
      "grad_norm": 3.9382618142315393,
      "learning_rate": 2.1582733812949642e-05,
      "loss": 3.2789,
      "step": 45
    },
    {
      "epoch": 0.0033157932674980177,
      "grad_norm": 4.442124676006116,
      "learning_rate": 2.2062350119904077e-05,
      "loss": 3.1227,
      "step": 46
    },
    {
      "epoch": 0.003387875729834931,
      "grad_norm": 3.4380288055703754,
      "learning_rate": 2.2541966426858512e-05,
      "loss": 3.3533,
      "step": 47
    },
    {
      "epoch": 0.003459958192171845,
      "grad_norm": 3.719192988036961,
      "learning_rate": 2.302158273381295e-05,
      "loss": 3.0983,
      "step": 48
    },
    {
      "epoch": 0.003532040654508758,
      "grad_norm": 3.6613251728039344,
      "learning_rate": 2.350119904076739e-05,
      "loss": 2.9367,
      "step": 49
    },
    {
      "epoch": 0.0036041231168456715,
      "grad_norm": 3.4468328225665545,
      "learning_rate": 2.3980815347721824e-05,
      "loss": 2.7623,
      "step": 50
    },
    {
      "epoch": 0.003676205579182585,
      "grad_norm": 3.6511233229972904,
      "learning_rate": 2.446043165467626e-05,
      "loss": 3.0131,
      "step": 51
    },
    {
      "epoch": 0.003748288041519498,
      "grad_norm": 3.2267490454309105,
      "learning_rate": 2.4940047961630698e-05,
      "loss": 2.9113,
      "step": 52
    },
    {
      "epoch": 0.003820370503856412,
      "grad_norm": 2.7663792807748493,
      "learning_rate": 2.541966426858513e-05,
      "loss": 2.7449,
      "step": 53
    },
    {
      "epoch": 0.0038924529661933252,
      "grad_norm": 3.6095058330597856,
      "learning_rate": 2.589928057553957e-05,
      "loss": 2.5675,
      "step": 54
    },
    {
      "epoch": 0.003964535428530239,
      "grad_norm": 2.682341378893519,
      "learning_rate": 2.637889688249401e-05,
      "loss": 2.807,
      "step": 55
    },
    {
      "epoch": 0.004036617890867152,
      "grad_norm": 3.829701749062023,
      "learning_rate": 2.6858513189448442e-05,
      "loss": 2.8209,
      "step": 56
    },
    {
      "epoch": 0.004108700353204066,
      "grad_norm": 3.635479035114784,
      "learning_rate": 2.733812949640288e-05,
      "loss": 2.7548,
      "step": 57
    },
    {
      "epoch": 0.004180782815540979,
      "grad_norm": 3.7096705271064327,
      "learning_rate": 2.7817745803357316e-05,
      "loss": 2.6394,
      "step": 58
    },
    {
      "epoch": 0.004252865277877892,
      "grad_norm": 4.307344260509464,
      "learning_rate": 2.8297362110311754e-05,
      "loss": 2.699,
      "step": 59
    },
    {
      "epoch": 0.004324947740214806,
      "grad_norm": 3.3802902054518897,
      "learning_rate": 2.8776978417266186e-05,
      "loss": 2.4621,
      "step": 60
    },
    {
      "epoch": 0.004397030202551719,
      "grad_norm": 2.964086444004209,
      "learning_rate": 2.9256594724220628e-05,
      "loss": 2.4325,
      "step": 61
    },
    {
      "epoch": 0.004469112664888633,
      "grad_norm": 2.434837979363635,
      "learning_rate": 2.973621103117506e-05,
      "loss": 2.3598,
      "step": 62
    },
    {
      "epoch": 0.004541195127225546,
      "grad_norm": 3.33638358916008,
      "learning_rate": 3.0215827338129498e-05,
      "loss": 2.7293,
      "step": 63
    },
    {
      "epoch": 0.0046132775895624594,
      "grad_norm": 2.9357782570043276,
      "learning_rate": 3.069544364508393e-05,
      "loss": 2.7048,
      "step": 64
    },
    {
      "epoch": 0.004685360051899373,
      "grad_norm": 2.293774134814458,
      "learning_rate": 3.117505995203837e-05,
      "loss": 2.2699,
      "step": 65
    },
    {
      "epoch": 0.004757442514236286,
      "grad_norm": 1.9520224969791389,
      "learning_rate": 3.165467625899281e-05,
      "loss": 2.3436,
      "step": 66
    },
    {
      "epoch": 0.0048295249765732,
      "grad_norm": 2.996718111519694,
      "learning_rate": 3.2134292565947245e-05,
      "loss": 2.2093,
      "step": 67
    },
    {
      "epoch": 0.004901607438910113,
      "grad_norm": 2.3939826447851393,
      "learning_rate": 3.2613908872901684e-05,
      "loss": 2.2851,
      "step": 68
    },
    {
      "epoch": 0.0049736899012470265,
      "grad_norm": 3.2838499754411647,
      "learning_rate": 3.3093525179856116e-05,
      "loss": 2.3213,
      "step": 69
    },
    {
      "epoch": 0.00504577236358394,
      "grad_norm": 2.8770570596820204,
      "learning_rate": 3.3573141486810554e-05,
      "loss": 2.3382,
      "step": 70
    },
    {
      "epoch": 0.005117854825920853,
      "grad_norm": 2.094106301890136,
      "learning_rate": 3.4052757793764986e-05,
      "loss": 2.2156,
      "step": 71
    },
    {
      "epoch": 0.005189937288257767,
      "grad_norm": 3.0231523693327977,
      "learning_rate": 3.453237410071943e-05,
      "loss": 2.2644,
      "step": 72
    },
    {
      "epoch": 0.00526201975059468,
      "grad_norm": 2.5520199351927673,
      "learning_rate": 3.501199040767386e-05,
      "loss": 2.1713,
      "step": 73
    },
    {
      "epoch": 0.005334102212931594,
      "grad_norm": 2.471672341000642,
      "learning_rate": 3.54916067146283e-05,
      "loss": 2.117,
      "step": 74
    },
    {
      "epoch": 0.005406184675268507,
      "grad_norm": 1.934306670049023,
      "learning_rate": 3.597122302158273e-05,
      "loss": 2.2116,
      "step": 75
    },
    {
      "epoch": 0.00547826713760542,
      "grad_norm": 2.1620831541971137,
      "learning_rate": 3.645083932853717e-05,
      "loss": 2.3549,
      "step": 76
    },
    {
      "epoch": 0.005550349599942334,
      "grad_norm": 2.109998314304401,
      "learning_rate": 3.6930455635491604e-05,
      "loss": 1.9709,
      "step": 77
    },
    {
      "epoch": 0.005622432062279248,
      "grad_norm": 1.7269689457748882,
      "learning_rate": 3.741007194244605e-05,
      "loss": 2.203,
      "step": 78
    },
    {
      "epoch": 0.005694514524616161,
      "grad_norm": 2.0502422591917866,
      "learning_rate": 3.788968824940048e-05,
      "loss": 2.0208,
      "step": 79
    },
    {
      "epoch": 0.0057665969869530745,
      "grad_norm": 1.806393215547253,
      "learning_rate": 3.836930455635492e-05,
      "loss": 2.028,
      "step": 80
    },
    {
      "epoch": 0.005838679449289987,
      "grad_norm": 2.3056536621434343,
      "learning_rate": 3.884892086330936e-05,
      "loss": 2.0358,
      "step": 81
    },
    {
      "epoch": 0.005910761911626901,
      "grad_norm": 1.4104070943533389,
      "learning_rate": 3.932853717026379e-05,
      "loss": 1.8764,
      "step": 82
    },
    {
      "epoch": 0.005982844373963815,
      "grad_norm": 2.1964405449716655,
      "learning_rate": 3.980815347721823e-05,
      "loss": 1.9779,
      "step": 83
    },
    {
      "epoch": 0.006054926836300728,
      "grad_norm": 2.5263302420987648,
      "learning_rate": 4.0287769784172666e-05,
      "loss": 2.0836,
      "step": 84
    },
    {
      "epoch": 0.006127009298637642,
      "grad_norm": 2.2305068428353327,
      "learning_rate": 4.0767386091127105e-05,
      "loss": 1.9206,
      "step": 85
    },
    {
      "epoch": 0.0061990917609745545,
      "grad_norm": 1.6345723498545293,
      "learning_rate": 4.124700239808154e-05,
      "loss": 2.0129,
      "step": 86
    },
    {
      "epoch": 0.006271174223311468,
      "grad_norm": 2.8774250722775703,
      "learning_rate": 4.1726618705035975e-05,
      "loss": 1.8738,
      "step": 87
    },
    {
      "epoch": 0.006343256685648382,
      "grad_norm": 1.4360390529911853,
      "learning_rate": 4.220623501199041e-05,
      "loss": 1.8698,
      "step": 88
    },
    {
      "epoch": 0.006415339147985295,
      "grad_norm": 1.982311359490484,
      "learning_rate": 4.2685851318944845e-05,
      "loss": 1.8838,
      "step": 89
    },
    {
      "epoch": 0.006487421610322209,
      "grad_norm": 1.4930657785229746,
      "learning_rate": 4.3165467625899284e-05,
      "loss": 1.8439,
      "step": 90
    },
    {
      "epoch": 0.006559504072659122,
      "grad_norm": 2.5242127439573947,
      "learning_rate": 4.364508393285372e-05,
      "loss": 1.9843,
      "step": 91
    },
    {
      "epoch": 0.006631586534996035,
      "grad_norm": 1.4438319538590543,
      "learning_rate": 4.4124700239808154e-05,
      "loss": 1.8731,
      "step": 92
    },
    {
      "epoch": 0.006703668997332949,
      "grad_norm": 1.7595239322773129,
      "learning_rate": 4.460431654676259e-05,
      "loss": 1.8105,
      "step": 93
    },
    {
      "epoch": 0.006775751459669862,
      "grad_norm": 2.188077458997556,
      "learning_rate": 4.5083932853717024e-05,
      "loss": 1.8948,
      "step": 94
    },
    {
      "epoch": 0.006847833922006776,
      "grad_norm": 1.268095122493525,
      "learning_rate": 4.556354916067146e-05,
      "loss": 1.8232,
      "step": 95
    },
    {
      "epoch": 0.00691991638434369,
      "grad_norm": 1.9342433231540643,
      "learning_rate": 4.60431654676259e-05,
      "loss": 2.0012,
      "step": 96
    },
    {
      "epoch": 0.0069919988466806025,
      "grad_norm": 1.320578036985148,
      "learning_rate": 4.652278177458034e-05,
      "loss": 1.5937,
      "step": 97
    },
    {
      "epoch": 0.007064081309017516,
      "grad_norm": 1.094084280294206,
      "learning_rate": 4.700239808153478e-05,
      "loss": 1.7267,
      "step": 98
    },
    {
      "epoch": 0.007136163771354429,
      "grad_norm": 1.9869485599132939,
      "learning_rate": 4.748201438848921e-05,
      "loss": 1.8557,
      "step": 99
    },
    {
      "epoch": 0.007208246233691343,
      "grad_norm": 2.0740222037705784,
      "learning_rate": 4.796163069544365e-05,
      "loss": 1.7105,
      "step": 100
    },
    {
      "epoch": 0.007280328696028257,
      "grad_norm": 1.7163109286020661,
      "learning_rate": 4.844124700239808e-05,
      "loss": 1.6405,
      "step": 101
    },
    {
      "epoch": 0.00735241115836517,
      "grad_norm": 2.7412573882070097,
      "learning_rate": 4.892086330935252e-05,
      "loss": 1.7995,
      "step": 102
    },
    {
      "epoch": 0.007424493620702083,
      "grad_norm": 2.319044403309231,
      "learning_rate": 4.940047961630696e-05,
      "loss": 1.8238,
      "step": 103
    },
    {
      "epoch": 0.007496576083038996,
      "grad_norm": 2.0579430311211895,
      "learning_rate": 4.9880095923261396e-05,
      "loss": 1.5964,
      "step": 104
    },
    {
      "epoch": 0.00756865854537591,
      "grad_norm": 1.8143403924404744,
      "learning_rate": 5.035971223021583e-05,
      "loss": 1.6925,
      "step": 105
    },
    {
      "epoch": 0.007640741007712824,
      "grad_norm": 1.854579007023678,
      "learning_rate": 5.083932853717026e-05,
      "loss": 1.83,
      "step": 106
    },
    {
      "epoch": 0.007712823470049737,
      "grad_norm": 2.2721812793296534,
      "learning_rate": 5.1318944844124705e-05,
      "loss": 1.6806,
      "step": 107
    },
    {
      "epoch": 0.0077849059323866505,
      "grad_norm": 1.9747143920109496,
      "learning_rate": 5.179856115107914e-05,
      "loss": 1.766,
      "step": 108
    },
    {
      "epoch": 0.007856988394723563,
      "grad_norm": 2.4830871966584884,
      "learning_rate": 5.2278177458033575e-05,
      "loss": 1.6509,
      "step": 109
    },
    {
      "epoch": 0.007929070857060478,
      "grad_norm": 1.2047310694903928,
      "learning_rate": 5.275779376498802e-05,
      "loss": 1.7866,
      "step": 110
    },
    {
      "epoch": 0.008001153319397391,
      "grad_norm": 2.174343373597398,
      "learning_rate": 5.323741007194245e-05,
      "loss": 1.7,
      "step": 111
    },
    {
      "epoch": 0.008073235781734304,
      "grad_norm": 1.144834276472731,
      "learning_rate": 5.3717026378896884e-05,
      "loss": 1.6155,
      "step": 112
    },
    {
      "epoch": 0.008145318244071217,
      "grad_norm": 2.7305608065850646,
      "learning_rate": 5.4196642685851316e-05,
      "loss": 1.6878,
      "step": 113
    },
    {
      "epoch": 0.008217400706408131,
      "grad_norm": 1.72373838097151,
      "learning_rate": 5.467625899280576e-05,
      "loss": 1.6753,
      "step": 114
    },
    {
      "epoch": 0.008289483168745044,
      "grad_norm": 2.5475984388091915,
      "learning_rate": 5.515587529976019e-05,
      "loss": 1.6798,
      "step": 115
    },
    {
      "epoch": 0.008361565631081957,
      "grad_norm": 2.390937820357757,
      "learning_rate": 5.563549160671463e-05,
      "loss": 1.6766,
      "step": 116
    },
    {
      "epoch": 0.008433648093418872,
      "grad_norm": 2.7662652868690896,
      "learning_rate": 5.611510791366906e-05,
      "loss": 1.638,
      "step": 117
    },
    {
      "epoch": 0.008505730555755785,
      "grad_norm": 2.312120198524236,
      "learning_rate": 5.659472422062351e-05,
      "loss": 1.6044,
      "step": 118
    },
    {
      "epoch": 0.008577813018092698,
      "grad_norm": 2.620198431452638,
      "learning_rate": 5.707434052757794e-05,
      "loss": 1.5488,
      "step": 119
    },
    {
      "epoch": 0.008649895480429612,
      "grad_norm": 2.270792424340099,
      "learning_rate": 5.755395683453237e-05,
      "loss": 1.7098,
      "step": 120
    },
    {
      "epoch": 0.008721977942766525,
      "grad_norm": 1.8166300638744977,
      "learning_rate": 5.803357314148681e-05,
      "loss": 1.6808,
      "step": 121
    },
    {
      "epoch": 0.008794060405103438,
      "grad_norm": 1.0218394212404056,
      "learning_rate": 5.8513189448441256e-05,
      "loss": 1.4651,
      "step": 122
    },
    {
      "epoch": 0.008866142867440351,
      "grad_norm": 1.2286772481417763,
      "learning_rate": 5.899280575539569e-05,
      "loss": 1.5214,
      "step": 123
    },
    {
      "epoch": 0.008938225329777266,
      "grad_norm": 1.4894438412503452,
      "learning_rate": 5.947242206235012e-05,
      "loss": 1.474,
      "step": 124
    },
    {
      "epoch": 0.009010307792114178,
      "grad_norm": 1.8161029670610813,
      "learning_rate": 5.9952038369304564e-05,
      "loss": 1.4911,
      "step": 125
    },
    {
      "epoch": 0.009082390254451091,
      "grad_norm": 2.1442043230976946,
      "learning_rate": 6.0431654676258996e-05,
      "loss": 1.5481,
      "step": 126
    },
    {
      "epoch": 0.009154472716788006,
      "grad_norm": 1.035850491962404,
      "learning_rate": 6.091127098321343e-05,
      "loss": 1.5143,
      "step": 127
    },
    {
      "epoch": 0.009226555179124919,
      "grad_norm": 0.8867122819646623,
      "learning_rate": 6.139088729016786e-05,
      "loss": 1.454,
      "step": 128
    },
    {
      "epoch": 0.009298637641461832,
      "grad_norm": 1.440248051573906,
      "learning_rate": 6.187050359712231e-05,
      "loss": 1.4695,
      "step": 129
    },
    {
      "epoch": 0.009370720103798746,
      "grad_norm": 0.7882900010137738,
      "learning_rate": 6.235011990407674e-05,
      "loss": 1.5388,
      "step": 130
    },
    {
      "epoch": 0.00944280256613566,
      "grad_norm": 0.8245902439861584,
      "learning_rate": 6.282973621103118e-05,
      "loss": 1.5226,
      "step": 131
    },
    {
      "epoch": 0.009514885028472572,
      "grad_norm": 1.0825196585351176,
      "learning_rate": 6.330935251798561e-05,
      "loss": 1.5395,
      "step": 132
    },
    {
      "epoch": 0.009586967490809487,
      "grad_norm": 1.1052568447248767,
      "learning_rate": 6.378896882494005e-05,
      "loss": 1.5083,
      "step": 133
    },
    {
      "epoch": 0.0096590499531464,
      "grad_norm": 2.0227781469950523,
      "learning_rate": 6.426858513189449e-05,
      "loss": 1.6185,
      "step": 134
    },
    {
      "epoch": 0.009731132415483313,
      "grad_norm": 1.2783367337727478,
      "learning_rate": 6.474820143884892e-05,
      "loss": 1.4865,
      "step": 135
    },
    {
      "epoch": 0.009803214877820226,
      "grad_norm": 1.7212813275633791,
      "learning_rate": 6.522781774580337e-05,
      "loss": 1.57,
      "step": 136
    },
    {
      "epoch": 0.00987529734015714,
      "grad_norm": 2.637873149831897,
      "learning_rate": 6.570743405275779e-05,
      "loss": 1.5033,
      "step": 137
    },
    {
      "epoch": 0.009947379802494053,
      "grad_norm": 0.7653615992789967,
      "learning_rate": 6.618705035971223e-05,
      "loss": 1.4717,
      "step": 138
    },
    {
      "epoch": 0.010019462264830966,
      "grad_norm": 2.572417366536391,
      "learning_rate": 6.666666666666667e-05,
      "loss": 1.4786,
      "step": 139
    },
    {
      "epoch": 0.01009154472716788,
      "grad_norm": 1.67112669707805,
      "learning_rate": 6.714628297362111e-05,
      "loss": 1.5309,
      "step": 140
    },
    {
      "epoch": 0.010163627189504794,
      "grad_norm": 0.6588302150300694,
      "learning_rate": 6.762589928057555e-05,
      "loss": 1.3542,
      "step": 141
    },
    {
      "epoch": 0.010235709651841706,
      "grad_norm": 0.6830499498465892,
      "learning_rate": 6.810551558752997e-05,
      "loss": 1.4295,
      "step": 142
    },
    {
      "epoch": 0.010307792114178621,
      "grad_norm": 0.8899174238046546,
      "learning_rate": 6.858513189448441e-05,
      "loss": 1.334,
      "step": 143
    },
    {
      "epoch": 0.010379874576515534,
      "grad_norm": 0.7661090698769075,
      "learning_rate": 6.906474820143886e-05,
      "loss": 1.418,
      "step": 144
    },
    {
      "epoch": 0.010451957038852447,
      "grad_norm": 0.8275744354655791,
      "learning_rate": 6.954436450839329e-05,
      "loss": 1.444,
      "step": 145
    },
    {
      "epoch": 0.01052403950118936,
      "grad_norm": 0.9659440334514485,
      "learning_rate": 7.002398081534773e-05,
      "loss": 1.4408,
      "step": 146
    },
    {
      "epoch": 0.010596121963526274,
      "grad_norm": 0.9777427687513432,
      "learning_rate": 7.050359712230215e-05,
      "loss": 1.3468,
      "step": 147
    },
    {
      "epoch": 0.010668204425863187,
      "grad_norm": 2.131786110750152,
      "learning_rate": 7.09832134292566e-05,
      "loss": 1.3704,
      "step": 148
    },
    {
      "epoch": 0.0107402868882001,
      "grad_norm": 1.0233636812834157,
      "learning_rate": 7.146282973621103e-05,
      "loss": 1.5317,
      "step": 149
    },
    {
      "epoch": 0.010812369350537015,
      "grad_norm": 1.1401288181135916,
      "learning_rate": 7.194244604316547e-05,
      "loss": 1.4209,
      "step": 150
    },
    {
      "epoch": 0.010884451812873928,
      "grad_norm": 1.5674192759369678,
      "learning_rate": 7.242206235011992e-05,
      "loss": 1.2962,
      "step": 151
    },
    {
      "epoch": 0.01095653427521084,
      "grad_norm": 1.2208573166261063,
      "learning_rate": 7.290167865707434e-05,
      "loss": 1.2698,
      "step": 152
    },
    {
      "epoch": 0.011028616737547755,
      "grad_norm": 1.3073583550337826,
      "learning_rate": 7.338129496402878e-05,
      "loss": 1.3932,
      "step": 153
    },
    {
      "epoch": 0.011100699199884668,
      "grad_norm": 0.6650942817917187,
      "learning_rate": 7.386091127098321e-05,
      "loss": 1.311,
      "step": 154
    },
    {
      "epoch": 0.011172781662221581,
      "grad_norm": 0.8024609681339686,
      "learning_rate": 7.434052757793766e-05,
      "loss": 1.3619,
      "step": 155
    },
    {
      "epoch": 0.011244864124558496,
      "grad_norm": 1.2538995273839437,
      "learning_rate": 7.48201438848921e-05,
      "loss": 1.3754,
      "step": 156
    },
    {
      "epoch": 0.011316946586895409,
      "grad_norm": 0.9333352538569224,
      "learning_rate": 7.529976019184652e-05,
      "loss": 1.3775,
      "step": 157
    },
    {
      "epoch": 0.011389029049232322,
      "grad_norm": 1.7991461885121356,
      "learning_rate": 7.577937649880096e-05,
      "loss": 1.4296,
      "step": 158
    },
    {
      "epoch": 0.011461111511569234,
      "grad_norm": 1.1886048046762239,
      "learning_rate": 7.62589928057554e-05,
      "loss": 1.4013,
      "step": 159
    },
    {
      "epoch": 0.011533193973906149,
      "grad_norm": 0.557112300546932,
      "learning_rate": 7.673860911270984e-05,
      "loss": 1.3131,
      "step": 160
    },
    {
      "epoch": 0.011605276436243062,
      "grad_norm": 0.5886531246267109,
      "learning_rate": 7.721822541966426e-05,
      "loss": 1.3707,
      "step": 161
    },
    {
      "epoch": 0.011677358898579975,
      "grad_norm": 0.5716057651733711,
      "learning_rate": 7.769784172661872e-05,
      "loss": 1.3588,
      "step": 162
    },
    {
      "epoch": 0.01174944136091689,
      "grad_norm": 1.3542378305609208,
      "learning_rate": 7.817745803357315e-05,
      "loss": 1.3816,
      "step": 163
    },
    {
      "epoch": 0.011821523823253802,
      "grad_norm": 1.1826477325436668,
      "learning_rate": 7.865707434052758e-05,
      "loss": 1.2617,
      "step": 164
    },
    {
      "epoch": 0.011893606285590715,
      "grad_norm": 1.0870495712081547,
      "learning_rate": 7.913669064748202e-05,
      "loss": 1.3853,
      "step": 165
    },
    {
      "epoch": 0.01196568874792763,
      "grad_norm": 1.287167518100364,
      "learning_rate": 7.961630695443646e-05,
      "loss": 1.435,
      "step": 166
    },
    {
      "epoch": 0.012037771210264543,
      "grad_norm": 1.8602365391546323,
      "learning_rate": 8.00959232613909e-05,
      "loss": 1.3686,
      "step": 167
    },
    {
      "epoch": 0.012109853672601456,
      "grad_norm": 1.2974156187531523,
      "learning_rate": 8.057553956834533e-05,
      "loss": 1.4858,
      "step": 168
    },
    {
      "epoch": 0.01218193613493837,
      "grad_norm": 0.4896175156664124,
      "learning_rate": 8.105515587529976e-05,
      "loss": 1.3713,
      "step": 169
    },
    {
      "epoch": 0.012254018597275283,
      "grad_norm": 2.1039737112027432,
      "learning_rate": 8.153477218225421e-05,
      "loss": 1.387,
      "step": 170
    },
    {
      "epoch": 0.012326101059612196,
      "grad_norm": 0.5461495867816092,
      "learning_rate": 8.201438848920863e-05,
      "loss": 1.2836,
      "step": 171
    },
    {
      "epoch": 0.012398183521949109,
      "grad_norm": 0.6226251115835553,
      "learning_rate": 8.249400479616307e-05,
      "loss": 1.3851,
      "step": 172
    },
    {
      "epoch": 0.012470265984286024,
      "grad_norm": 0.5979785000991408,
      "learning_rate": 8.29736211031175e-05,
      "loss": 1.3737,
      "step": 173
    },
    {
      "epoch": 0.012542348446622937,
      "grad_norm": 2.0464789951501157,
      "learning_rate": 8.345323741007195e-05,
      "loss": 1.3751,
      "step": 174
    },
    {
      "epoch": 0.01261443090895985,
      "grad_norm": 0.5519754858206654,
      "learning_rate": 8.393285371702639e-05,
      "loss": 1.3543,
      "step": 175
    },
    {
      "epoch": 0.012686513371296764,
      "grad_norm": 1.7454522980150025,
      "learning_rate": 8.441247002398081e-05,
      "loss": 1.4247,
      "step": 176
    },
    {
      "epoch": 0.012758595833633677,
      "grad_norm": 0.6456311900459133,
      "learning_rate": 8.489208633093527e-05,
      "loss": 1.2658,
      "step": 177
    },
    {
      "epoch": 0.01283067829597059,
      "grad_norm": 1.029130722588737,
      "learning_rate": 8.537170263788969e-05,
      "loss": 1.3857,
      "step": 178
    },
    {
      "epoch": 0.012902760758307505,
      "grad_norm": 0.7404589773762769,
      "learning_rate": 8.585131894484413e-05,
      "loss": 1.298,
      "step": 179
    },
    {
      "epoch": 0.012974843220644417,
      "grad_norm": 1.3769510909575233,
      "learning_rate": 8.633093525179857e-05,
      "loss": 1.3591,
      "step": 180
    },
    {
      "epoch": 0.01304692568298133,
      "grad_norm": 0.7051135056897345,
      "learning_rate": 8.6810551558753e-05,
      "loss": 1.327,
      "step": 181
    },
    {
      "epoch": 0.013119008145318243,
      "grad_norm": 1.3155478591747336,
      "learning_rate": 8.729016786570744e-05,
      "loss": 1.4087,
      "step": 182
    },
    {
      "epoch": 0.013191090607655158,
      "grad_norm": 0.8116081194386942,
      "learning_rate": 8.776978417266187e-05,
      "loss": 1.3518,
      "step": 183
    },
    {
      "epoch": 0.01326317306999207,
      "grad_norm": 0.5716153683641211,
      "learning_rate": 8.824940047961631e-05,
      "loss": 1.2258,
      "step": 184
    },
    {
      "epoch": 0.013335255532328984,
      "grad_norm": 0.750108430871394,
      "learning_rate": 8.872901678657075e-05,
      "loss": 1.3649,
      "step": 185
    },
    {
      "epoch": 0.013407337994665898,
      "grad_norm": 0.6106107762495203,
      "learning_rate": 8.920863309352519e-05,
      "loss": 1.2809,
      "step": 186
    },
    {
      "epoch": 0.013479420457002811,
      "grad_norm": 1.3791265273470623,
      "learning_rate": 8.968824940047962e-05,
      "loss": 1.2497,
      "step": 187
    },
    {
      "epoch": 0.013551502919339724,
      "grad_norm": 0.6884792873266674,
      "learning_rate": 9.016786570743405e-05,
      "loss": 1.2759,
      "step": 188
    },
    {
      "epoch": 0.013623585381676639,
      "grad_norm": 0.949882669413638,
      "learning_rate": 9.06474820143885e-05,
      "loss": 1.2821,
      "step": 189
    },
    {
      "epoch": 0.013695667844013552,
      "grad_norm": 0.7276352301729418,
      "learning_rate": 9.112709832134293e-05,
      "loss": 1.2858,
      "step": 190
    },
    {
      "epoch": 0.013767750306350465,
      "grad_norm": 1.0736182319947143,
      "learning_rate": 9.160671462829736e-05,
      "loss": 1.2896,
      "step": 191
    },
    {
      "epoch": 0.01383983276868738,
      "grad_norm": 0.8045994243639462,
      "learning_rate": 9.20863309352518e-05,
      "loss": 1.3548,
      "step": 192
    },
    {
      "epoch": 0.013911915231024292,
      "grad_norm": 1.9792500482663749,
      "learning_rate": 9.256594724220624e-05,
      "loss": 1.2913,
      "step": 193
    },
    {
      "epoch": 0.013983997693361205,
      "grad_norm": 0.9207462108739795,
      "learning_rate": 9.304556354916068e-05,
      "loss": 1.2247,
      "step": 194
    },
    {
      "epoch": 0.014056080155698118,
      "grad_norm": 1.0689309438107852,
      "learning_rate": 9.35251798561151e-05,
      "loss": 1.3059,
      "step": 195
    },
    {
      "epoch": 0.014128162618035033,
      "grad_norm": 0.4400220531448739,
      "learning_rate": 9.400479616306956e-05,
      "loss": 1.2311,
      "step": 196
    },
    {
      "epoch": 0.014200245080371945,
      "grad_norm": 0.719592476680441,
      "learning_rate": 9.448441247002398e-05,
      "loss": 1.2089,
      "step": 197
    },
    {
      "epoch": 0.014272327542708858,
      "grad_norm": 1.4842828988848722,
      "learning_rate": 9.496402877697842e-05,
      "loss": 1.1809,
      "step": 198
    },
    {
      "epoch": 0.014344410005045773,
      "grad_norm": 0.5798037769503822,
      "learning_rate": 9.544364508393286e-05,
      "loss": 1.271,
      "step": 199
    },
    {
      "epoch": 0.014416492467382686,
      "grad_norm": 1.1965349357945947,
      "learning_rate": 9.59232613908873e-05,
      "loss": 1.3672,
      "step": 200
    },
    {
      "epoch": 0.014488574929719599,
      "grad_norm": 0.6507268824036464,
      "learning_rate": 9.640287769784174e-05,
      "loss": 1.3523,
      "step": 201
    },
    {
      "epoch": 0.014560657392056513,
      "grad_norm": 0.6998396849111597,
      "learning_rate": 9.688249400479616e-05,
      "loss": 1.3324,
      "step": 202
    },
    {
      "epoch": 0.014632739854393426,
      "grad_norm": 0.6359719237548558,
      "learning_rate": 9.736211031175061e-05,
      "loss": 1.2887,
      "step": 203
    },
    {
      "epoch": 0.01470482231673034,
      "grad_norm": 0.5324526887330234,
      "learning_rate": 9.784172661870504e-05,
      "loss": 1.2569,
      "step": 204
    },
    {
      "epoch": 0.014776904779067252,
      "grad_norm": 0.7108483436654904,
      "learning_rate": 9.832134292565948e-05,
      "loss": 1.3344,
      "step": 205
    },
    {
      "epoch": 0.014848987241404167,
      "grad_norm": 2.1962198767098227,
      "learning_rate": 9.880095923261392e-05,
      "loss": 1.3688,
      "step": 206
    },
    {
      "epoch": 0.01492106970374108,
      "grad_norm": 0.7216884060074366,
      "learning_rate": 9.928057553956835e-05,
      "loss": 1.3293,
      "step": 207
    },
    {
      "epoch": 0.014993152166077993,
      "grad_norm": 3.1409457343607503,
      "learning_rate": 9.976019184652279e-05,
      "loss": 1.3539,
      "step": 208
    },
    {
      "epoch": 0.015065234628414907,
      "grad_norm": 1.9761696998136127,
      "learning_rate": 0.00010023980815347722,
      "loss": 1.1954,
      "step": 209
    },
    {
      "epoch": 0.01513731709075182,
      "grad_norm": 0.6531068064395966,
      "learning_rate": 0.00010071942446043166,
      "loss": 1.3465,
      "step": 210
    },
    {
      "epoch": 0.015209399553088733,
      "grad_norm": 1.7246044224494659,
      "learning_rate": 0.0001011990407673861,
      "loss": 1.2946,
      "step": 211
    },
    {
      "epoch": 0.015281482015425648,
      "grad_norm": 0.5527332895685734,
      "learning_rate": 0.00010167865707434052,
      "loss": 1.251,
      "step": 212
    },
    {
      "epoch": 0.01535356447776256,
      "grad_norm": 1.4233808966187624,
      "learning_rate": 0.00010215827338129497,
      "loss": 1.3015,
      "step": 213
    },
    {
      "epoch": 0.015425646940099473,
      "grad_norm": 0.507759555690241,
      "learning_rate": 0.00010263788968824941,
      "loss": 1.2461,
      "step": 214
    },
    {
      "epoch": 0.015497729402436388,
      "grad_norm": 1.1877079826402284,
      "learning_rate": 0.00010311750599520385,
      "loss": 1.1858,
      "step": 215
    },
    {
      "epoch": 0.015569811864773301,
      "grad_norm": 0.7209217312077615,
      "learning_rate": 0.00010359712230215827,
      "loss": 1.4265,
      "step": 216
    },
    {
      "epoch": 0.015641894327110214,
      "grad_norm": 0.8211483067121337,
      "learning_rate": 0.00010407673860911271,
      "loss": 1.2582,
      "step": 217
    },
    {
      "epoch": 0.015713976789447127,
      "grad_norm": 0.6243628549096882,
      "learning_rate": 0.00010455635491606715,
      "loss": 1.2642,
      "step": 218
    },
    {
      "epoch": 0.01578605925178404,
      "grad_norm": 1.8256128606682935,
      "learning_rate": 0.00010503597122302158,
      "loss": 1.4281,
      "step": 219
    },
    {
      "epoch": 0.015858141714120956,
      "grad_norm": 0.5151939453284471,
      "learning_rate": 0.00010551558752997604,
      "loss": 1.3108,
      "step": 220
    },
    {
      "epoch": 0.01593022417645787,
      "grad_norm": 0.6650184171742007,
      "learning_rate": 0.00010599520383693047,
      "loss": 1.2841,
      "step": 221
    },
    {
      "epoch": 0.016002306638794782,
      "grad_norm": 0.9187048494186336,
      "learning_rate": 0.0001064748201438849,
      "loss": 1.2283,
      "step": 222
    },
    {
      "epoch": 0.016074389101131695,
      "grad_norm": 1.4871728758016096,
      "learning_rate": 0.00010695443645083933,
      "loss": 1.2348,
      "step": 223
    },
    {
      "epoch": 0.016146471563468608,
      "grad_norm": 1.5686135663912304,
      "learning_rate": 0.00010743405275779377,
      "loss": 1.1767,
      "step": 224
    },
    {
      "epoch": 0.01621855402580552,
      "grad_norm": 0.5960380835663105,
      "learning_rate": 0.0001079136690647482,
      "loss": 1.3554,
      "step": 225
    },
    {
      "epoch": 0.016290636488142433,
      "grad_norm": 1.2528649804462215,
      "learning_rate": 0.00010839328537170263,
      "loss": 1.252,
      "step": 226
    },
    {
      "epoch": 0.01636271895047935,
      "grad_norm": 2.118692948203393,
      "learning_rate": 0.00010887290167865707,
      "loss": 1.2865,
      "step": 227
    },
    {
      "epoch": 0.016434801412816263,
      "grad_norm": 0.9564024295481147,
      "learning_rate": 0.00010935251798561152,
      "loss": 1.2195,
      "step": 228
    },
    {
      "epoch": 0.016506883875153176,
      "grad_norm": 0.7946891926096098,
      "learning_rate": 0.00010983213429256596,
      "loss": 1.2649,
      "step": 229
    },
    {
      "epoch": 0.01657896633749009,
      "grad_norm": 0.7981739573292709,
      "learning_rate": 0.00011031175059952039,
      "loss": 1.223,
      "step": 230
    },
    {
      "epoch": 0.016651048799827,
      "grad_norm": 0.9199190555227229,
      "learning_rate": 0.00011079136690647482,
      "loss": 1.2456,
      "step": 231
    },
    {
      "epoch": 0.016723131262163914,
      "grad_norm": 1.095405994195379,
      "learning_rate": 0.00011127098321342926,
      "loss": 1.2189,
      "step": 232
    },
    {
      "epoch": 0.01679521372450083,
      "grad_norm": 0.5026155335550127,
      "learning_rate": 0.00011175059952038369,
      "loss": 1.2664,
      "step": 233
    },
    {
      "epoch": 0.016867296186837744,
      "grad_norm": 0.5554944876664324,
      "learning_rate": 0.00011223021582733813,
      "loss": 1.1938,
      "step": 234
    },
    {
      "epoch": 0.016939378649174656,
      "grad_norm": 0.9676421011742996,
      "learning_rate": 0.00011270983213429258,
      "loss": 1.278,
      "step": 235
    },
    {
      "epoch": 0.01701146111151157,
      "grad_norm": 1.279826465970854,
      "learning_rate": 0.00011318944844124702,
      "loss": 1.2859,
      "step": 236
    },
    {
      "epoch": 0.017083543573848482,
      "grad_norm": 0.5784298790658813,
      "learning_rate": 0.00011366906474820144,
      "loss": 1.2687,
      "step": 237
    },
    {
      "epoch": 0.017155626036185395,
      "grad_norm": 0.40992951720665716,
      "learning_rate": 0.00011414868105515588,
      "loss": 1.3019,
      "step": 238
    },
    {
      "epoch": 0.017227708498522308,
      "grad_norm": 0.3961701600207948,
      "learning_rate": 0.00011462829736211032,
      "loss": 1.3095,
      "step": 239
    },
    {
      "epoch": 0.017299790960859224,
      "grad_norm": 0.5301929436035578,
      "learning_rate": 0.00011510791366906474,
      "loss": 1.2757,
      "step": 240
    },
    {
      "epoch": 0.017371873423196137,
      "grad_norm": 0.9693289693862235,
      "learning_rate": 0.00011558752997601918,
      "loss": 1.2621,
      "step": 241
    },
    {
      "epoch": 0.01744395588553305,
      "grad_norm": 0.6243475949741931,
      "learning_rate": 0.00011606714628297362,
      "loss": 1.2782,
      "step": 242
    },
    {
      "epoch": 0.017516038347869963,
      "grad_norm": 0.4931490492198597,
      "learning_rate": 0.00011654676258992807,
      "loss": 1.2409,
      "step": 243
    },
    {
      "epoch": 0.017588120810206876,
      "grad_norm": 0.49321346082754547,
      "learning_rate": 0.00011702637889688251,
      "loss": 1.1257,
      "step": 244
    },
    {
      "epoch": 0.01766020327254379,
      "grad_norm": 0.7553520596852796,
      "learning_rate": 0.00011750599520383694,
      "loss": 1.2808,
      "step": 245
    },
    {
      "epoch": 0.017732285734880702,
      "grad_norm": 0.4074587855430615,
      "learning_rate": 0.00011798561151079137,
      "loss": 1.1477,
      "step": 246
    },
    {
      "epoch": 0.017804368197217618,
      "grad_norm": 0.5319034430109638,
      "learning_rate": 0.0001184652278177458,
      "loss": 1.2518,
      "step": 247
    },
    {
      "epoch": 0.01787645065955453,
      "grad_norm": 1.5447012291996596,
      "learning_rate": 0.00011894484412470024,
      "loss": 1.2586,
      "step": 248
    },
    {
      "epoch": 0.017948533121891444,
      "grad_norm": 0.5524478661732433,
      "learning_rate": 0.00011942446043165468,
      "loss": 1.2401,
      "step": 249
    },
    {
      "epoch": 0.018020615584228357,
      "grad_norm": 0.40347338812620287,
      "learning_rate": 0.00011990407673860913,
      "loss": 1.2298,
      "step": 250
    },
    {
      "epoch": 0.01809269804656527,
      "grad_norm": 0.5448320346425857,
      "learning_rate": 0.00012038369304556357,
      "loss": 1.2505,
      "step": 251
    },
    {
      "epoch": 0.018164780508902183,
      "grad_norm": 0.797969638898525,
      "learning_rate": 0.00012086330935251799,
      "loss": 1.3337,
      "step": 252
    },
    {
      "epoch": 0.0182368629712391,
      "grad_norm": 0.38376949285209094,
      "learning_rate": 0.00012134292565947243,
      "loss": 1.1817,
      "step": 253
    },
    {
      "epoch": 0.018308945433576012,
      "grad_norm": 1.7151548368627012,
      "learning_rate": 0.00012182254196642686,
      "loss": 1.1636,
      "step": 254
    },
    {
      "epoch": 0.018381027895912925,
      "grad_norm": 0.3819141354963388,
      "learning_rate": 0.0001223021582733813,
      "loss": 1.2693,
      "step": 255
    },
    {
      "epoch": 0.018453110358249838,
      "grad_norm": 0.38426451567791614,
      "learning_rate": 0.00012278177458033572,
      "loss": 1.1711,
      "step": 256
    },
    {
      "epoch": 0.01852519282058675,
      "grad_norm": 0.8255728439680214,
      "learning_rate": 0.00012326139088729017,
      "loss": 1.1811,
      "step": 257
    },
    {
      "epoch": 0.018597275282923664,
      "grad_norm": 0.46460978301070865,
      "learning_rate": 0.00012374100719424462,
      "loss": 1.2269,
      "step": 258
    },
    {
      "epoch": 0.018669357745260576,
      "grad_norm": 1.2942039860589283,
      "learning_rate": 0.00012422062350119905,
      "loss": 1.2903,
      "step": 259
    },
    {
      "epoch": 0.018741440207597493,
      "grad_norm": 0.8889318105321089,
      "learning_rate": 0.00012470023980815347,
      "loss": 1.2007,
      "step": 260
    },
    {
      "epoch": 0.018813522669934406,
      "grad_norm": 2.232124287170919,
      "learning_rate": 0.00012517985611510793,
      "loss": 1.3003,
      "step": 261
    },
    {
      "epoch": 0.01888560513227132,
      "grad_norm": 0.45275849753551267,
      "learning_rate": 0.00012565947242206235,
      "loss": 1.1704,
      "step": 262
    },
    {
      "epoch": 0.01895768759460823,
      "grad_norm": 1.5332568212872064,
      "learning_rate": 0.00012613908872901678,
      "loss": 1.1993,
      "step": 263
    },
    {
      "epoch": 0.019029770056945144,
      "grad_norm": 0.7142981761702881,
      "learning_rate": 0.00012661870503597123,
      "loss": 1.2849,
      "step": 264
    },
    {
      "epoch": 0.019101852519282057,
      "grad_norm": 0.3287014969013025,
      "learning_rate": 0.00012709832134292568,
      "loss": 1.1648,
      "step": 265
    },
    {
      "epoch": 0.019173934981618974,
      "grad_norm": 1.6797090886767105,
      "learning_rate": 0.0001275779376498801,
      "loss": 1.2543,
      "step": 266
    },
    {
      "epoch": 0.019246017443955887,
      "grad_norm": 0.6031404549777287,
      "learning_rate": 0.00012805755395683453,
      "loss": 1.2216,
      "step": 267
    },
    {
      "epoch": 0.0193180999062928,
      "grad_norm": 0.3068671303058401,
      "learning_rate": 0.00012853717026378898,
      "loss": 1.1985,
      "step": 268
    },
    {
      "epoch": 0.019390182368629712,
      "grad_norm": 0.5740465177808278,
      "learning_rate": 0.0001290167865707434,
      "loss": 1.1919,
      "step": 269
    },
    {
      "epoch": 0.019462264830966625,
      "grad_norm": 1.0334027066644715,
      "learning_rate": 0.00012949640287769783,
      "loss": 1.1521,
      "step": 270
    },
    {
      "epoch": 0.019534347293303538,
      "grad_norm": 0.47924660603138824,
      "learning_rate": 0.00012997601918465228,
      "loss": 1.2818,
      "step": 271
    },
    {
      "epoch": 0.01960642975564045,
      "grad_norm": 0.5120328954010321,
      "learning_rate": 0.00013045563549160674,
      "loss": 1.2262,
      "step": 272
    },
    {
      "epoch": 0.019678512217977367,
      "grad_norm": 1.1213029068423401,
      "learning_rate": 0.00013093525179856116,
      "loss": 1.2484,
      "step": 273
    },
    {
      "epoch": 0.01975059468031428,
      "grad_norm": 0.6374149433212346,
      "learning_rate": 0.00013141486810551559,
      "loss": 1.2663,
      "step": 274
    },
    {
      "epoch": 0.019822677142651193,
      "grad_norm": 0.9015063042859868,
      "learning_rate": 0.00013189448441247004,
      "loss": 1.1907,
      "step": 275
    },
    {
      "epoch": 0.019894759604988106,
      "grad_norm": 0.5618325266993317,
      "learning_rate": 0.00013237410071942446,
      "loss": 1.236,
      "step": 276
    },
    {
      "epoch": 0.01996684206732502,
      "grad_norm": 0.5884440170174238,
      "learning_rate": 0.0001328537170263789,
      "loss": 1.2681,
      "step": 277
    },
    {
      "epoch": 0.020038924529661932,
      "grad_norm": 0.9145726493673046,
      "learning_rate": 0.00013333333333333334,
      "loss": 1.2697,
      "step": 278
    },
    {
      "epoch": 0.02011100699199885,
      "grad_norm": 0.5915349214827341,
      "learning_rate": 0.00013381294964028776,
      "loss": 1.2666,
      "step": 279
    },
    {
      "epoch": 0.02018308945433576,
      "grad_norm": 0.5459688711683621,
      "learning_rate": 0.00013429256594724222,
      "loss": 1.1293,
      "step": 280
    },
    {
      "epoch": 0.020255171916672674,
      "grad_norm": 0.5383836314485797,
      "learning_rate": 0.00013477218225419664,
      "loss": 1.1496,
      "step": 281
    },
    {
      "epoch": 0.020327254379009587,
      "grad_norm": 1.447041643029839,
      "learning_rate": 0.0001352517985611511,
      "loss": 1.3014,
      "step": 282
    },
    {
      "epoch": 0.0203993368413465,
      "grad_norm": 0.31670096953491605,
      "learning_rate": 0.00013573141486810552,
      "loss": 1.28,
      "step": 283
    },
    {
      "epoch": 0.020471419303683413,
      "grad_norm": 0.9277033083236227,
      "learning_rate": 0.00013621103117505994,
      "loss": 1.2759,
      "step": 284
    },
    {
      "epoch": 0.020543501766020326,
      "grad_norm": 0.2796662640293699,
      "learning_rate": 0.0001366906474820144,
      "loss": 1.1884,
      "step": 285
    },
    {
      "epoch": 0.020615584228357242,
      "grad_norm": 0.31256008103938737,
      "learning_rate": 0.00013717026378896882,
      "loss": 1.2329,
      "step": 286
    },
    {
      "epoch": 0.020687666690694155,
      "grad_norm": 0.47074321865810664,
      "learning_rate": 0.00013764988009592327,
      "loss": 1.1936,
      "step": 287
    },
    {
      "epoch": 0.020759749153031068,
      "grad_norm": 1.0066757399373474,
      "learning_rate": 0.00013812949640287772,
      "loss": 1.2199,
      "step": 288
    },
    {
      "epoch": 0.02083183161536798,
      "grad_norm": 0.4761582061015232,
      "learning_rate": 0.00013860911270983215,
      "loss": 1.1729,
      "step": 289
    },
    {
      "epoch": 0.020903914077704894,
      "grad_norm": 1.7034886575060881,
      "learning_rate": 0.00013908872901678657,
      "loss": 1.2411,
      "step": 290
    },
    {
      "epoch": 0.020975996540041807,
      "grad_norm": 0.31302599317320773,
      "learning_rate": 0.000139568345323741,
      "loss": 1.1558,
      "step": 291
    },
    {
      "epoch": 0.02104807900237872,
      "grad_norm": 1.7931178951708207,
      "learning_rate": 0.00014004796163069545,
      "loss": 1.21,
      "step": 292
    },
    {
      "epoch": 0.021120161464715636,
      "grad_norm": 0.4509885662868183,
      "learning_rate": 0.00014052757793764988,
      "loss": 1.2082,
      "step": 293
    },
    {
      "epoch": 0.02119224392705255,
      "grad_norm": 0.775862527035015,
      "learning_rate": 0.0001410071942446043,
      "loss": 1.2306,
      "step": 294
    },
    {
      "epoch": 0.02126432638938946,
      "grad_norm": 0.5152607521244243,
      "learning_rate": 0.00014148681055155878,
      "loss": 1.175,
      "step": 295
    },
    {
      "epoch": 0.021336408851726375,
      "grad_norm": 0.36592585750130907,
      "learning_rate": 0.0001419664268585132,
      "loss": 1.2858,
      "step": 296
    },
    {
      "epoch": 0.021408491314063288,
      "grad_norm": 0.5984237350681603,
      "learning_rate": 0.00014244604316546763,
      "loss": 1.304,
      "step": 297
    },
    {
      "epoch": 0.0214805737764002,
      "grad_norm": 0.32862799251531116,
      "learning_rate": 0.00014292565947242206,
      "loss": 1.1064,
      "step": 298
    },
    {
      "epoch": 0.021552656238737117,
      "grad_norm": 1.2311278546075002,
      "learning_rate": 0.0001434052757793765,
      "loss": 1.2071,
      "step": 299
    },
    {
      "epoch": 0.02162473870107403,
      "grad_norm": 0.5825660214813013,
      "learning_rate": 0.00014388489208633093,
      "loss": 1.205,
      "step": 300
    },
    {
      "epoch": 0.021696821163410943,
      "grad_norm": 0.5747468006841465,
      "learning_rate": 0.00014436450839328536,
      "loss": 1.2053,
      "step": 301
    },
    {
      "epoch": 0.021768903625747855,
      "grad_norm": 0.5287359969520349,
      "learning_rate": 0.00014484412470023984,
      "loss": 1.1574,
      "step": 302
    },
    {
      "epoch": 0.02184098608808477,
      "grad_norm": 1.4684568005187184,
      "learning_rate": 0.00014532374100719426,
      "loss": 1.2296,
      "step": 303
    },
    {
      "epoch": 0.02191306855042168,
      "grad_norm": 1.2283735264952111,
      "learning_rate": 0.0001458033573141487,
      "loss": 1.1675,
      "step": 304
    },
    {
      "epoch": 0.021985151012758594,
      "grad_norm": 0.7517462719826137,
      "learning_rate": 0.0001462829736211031,
      "loss": 1.19,
      "step": 305
    },
    {
      "epoch": 0.02205723347509551,
      "grad_norm": 0.5900491318107791,
      "learning_rate": 0.00014676258992805756,
      "loss": 1.1424,
      "step": 306
    },
    {
      "epoch": 0.022129315937432423,
      "grad_norm": 0.3205699675879301,
      "learning_rate": 0.000147242206235012,
      "loss": 1.2412,
      "step": 307
    },
    {
      "epoch": 0.022201398399769336,
      "grad_norm": 0.3926245594704196,
      "learning_rate": 0.00014772182254196641,
      "loss": 1.2213,
      "step": 308
    },
    {
      "epoch": 0.02227348086210625,
      "grad_norm": 0.4566368014863002,
      "learning_rate": 0.00014820143884892087,
      "loss": 1.2357,
      "step": 309
    },
    {
      "epoch": 0.022345563324443162,
      "grad_norm": 0.5326636488822529,
      "learning_rate": 0.00014868105515587532,
      "loss": 1.2477,
      "step": 310
    },
    {
      "epoch": 0.022417645786780075,
      "grad_norm": 0.48573791657039617,
      "learning_rate": 0.00014916067146282974,
      "loss": 1.2464,
      "step": 311
    },
    {
      "epoch": 0.02248972824911699,
      "grad_norm": 0.34353847425598993,
      "learning_rate": 0.0001496402877697842,
      "loss": 1.1869,
      "step": 312
    },
    {
      "epoch": 0.022561810711453904,
      "grad_norm": 0.27533883227669775,
      "learning_rate": 0.00015011990407673862,
      "loss": 1.1692,
      "step": 313
    },
    {
      "epoch": 0.022633893173790817,
      "grad_norm": 0.7981622353807449,
      "learning_rate": 0.00015059952038369305,
      "loss": 1.1719,
      "step": 314
    },
    {
      "epoch": 0.02270597563612773,
      "grad_norm": 0.49695310894132033,
      "learning_rate": 0.00015107913669064747,
      "loss": 1.1568,
      "step": 315
    },
    {
      "epoch": 0.022778058098464643,
      "grad_norm": 0.8430673826258722,
      "learning_rate": 0.00015155875299760192,
      "loss": 1.316,
      "step": 316
    },
    {
      "epoch": 0.022850140560801556,
      "grad_norm": 0.36208176211893367,
      "learning_rate": 0.00015203836930455637,
      "loss": 1.1762,
      "step": 317
    },
    {
      "epoch": 0.02292222302313847,
      "grad_norm": 0.5644263507831551,
      "learning_rate": 0.0001525179856115108,
      "loss": 1.265,
      "step": 318
    },
    {
      "epoch": 0.022994305485475385,
      "grad_norm": 0.5853956258809863,
      "learning_rate": 0.00015299760191846525,
      "loss": 1.2352,
      "step": 319
    },
    {
      "epoch": 0.023066387947812298,
      "grad_norm": 0.3234220494764489,
      "learning_rate": 0.00015347721822541968,
      "loss": 1.1469,
      "step": 320
    },
    {
      "epoch": 0.02313847041014921,
      "grad_norm": 1.95182852127497,
      "learning_rate": 0.0001539568345323741,
      "loss": 1.2972,
      "step": 321
    },
    {
      "epoch": 0.023210552872486124,
      "grad_norm": 0.872275322556988,
      "learning_rate": 0.00015443645083932853,
      "loss": 1.1423,
      "step": 322
    },
    {
      "epoch": 0.023282635334823037,
      "grad_norm": 0.4922069222548833,
      "learning_rate": 0.00015491606714628298,
      "loss": 1.2012,
      "step": 323
    },
    {
      "epoch": 0.02335471779715995,
      "grad_norm": 0.6081407191408731,
      "learning_rate": 0.00015539568345323743,
      "loss": 1.2304,
      "step": 324
    },
    {
      "epoch": 0.023426800259496866,
      "grad_norm": 0.6837990051298278,
      "learning_rate": 0.00015587529976019186,
      "loss": 1.1681,
      "step": 325
    },
    {
      "epoch": 0.02349888272183378,
      "grad_norm": 0.38357052158167154,
      "learning_rate": 0.0001563549160671463,
      "loss": 1.2052,
      "step": 326
    },
    {
      "epoch": 0.023570965184170692,
      "grad_norm": 0.8525930126319451,
      "learning_rate": 0.00015683453237410073,
      "loss": 1.1814,
      "step": 327
    },
    {
      "epoch": 0.023643047646507605,
      "grad_norm": 0.3207733594845039,
      "learning_rate": 0.00015731414868105516,
      "loss": 1.1969,
      "step": 328
    },
    {
      "epoch": 0.023715130108844518,
      "grad_norm": 0.4751321785977216,
      "learning_rate": 0.00015779376498800958,
      "loss": 1.1713,
      "step": 329
    },
    {
      "epoch": 0.02378721257118143,
      "grad_norm": 1.9581138493550736,
      "learning_rate": 0.00015827338129496403,
      "loss": 1.1641,
      "step": 330
    },
    {
      "epoch": 0.023859295033518343,
      "grad_norm": 0.39809082128862466,
      "learning_rate": 0.00015875299760191846,
      "loss": 1.1843,
      "step": 331
    },
    {
      "epoch": 0.02393137749585526,
      "grad_norm": 0.3979162691250705,
      "learning_rate": 0.0001592326139088729,
      "loss": 1.1647,
      "step": 332
    },
    {
      "epoch": 0.024003459958192173,
      "grad_norm": 0.34822692274950284,
      "learning_rate": 0.00015971223021582736,
      "loss": 1.1736,
      "step": 333
    },
    {
      "epoch": 0.024075542420529086,
      "grad_norm": 0.30331544878401684,
      "learning_rate": 0.0001601918465227818,
      "loss": 1.1669,
      "step": 334
    },
    {
      "epoch": 0.024147624882866,
      "grad_norm": 1.3192821518606122,
      "learning_rate": 0.0001606714628297362,
      "loss": 1.1382,
      "step": 335
    },
    {
      "epoch": 0.02421970734520291,
      "grad_norm": 0.295331841241905,
      "learning_rate": 0.00016115107913669067,
      "loss": 1.1146,
      "step": 336
    },
    {
      "epoch": 0.024291789807539824,
      "grad_norm": 1.3228511517851855,
      "learning_rate": 0.0001616306954436451,
      "loss": 1.1794,
      "step": 337
    },
    {
      "epoch": 0.02436387226987674,
      "grad_norm": 1.2726349422041554,
      "learning_rate": 0.00016211031175059952,
      "loss": 1.2237,
      "step": 338
    },
    {
      "epoch": 0.024435954732213654,
      "grad_norm": 0.30251520315372404,
      "learning_rate": 0.00016258992805755397,
      "loss": 1.1846,
      "step": 339
    },
    {
      "epoch": 0.024508037194550567,
      "grad_norm": 1.2683039709967605,
      "learning_rate": 0.00016306954436450842,
      "loss": 1.224,
      "step": 340
    },
    {
      "epoch": 0.02458011965688748,
      "grad_norm": 1.2435863121474802,
      "learning_rate": 0.00016354916067146284,
      "loss": 1.2748,
      "step": 341
    },
    {
      "epoch": 0.024652202119224392,
      "grad_norm": 0.954788539332174,
      "learning_rate": 0.00016402877697841727,
      "loss": 1.2106,
      "step": 342
    },
    {
      "epoch": 0.024724284581561305,
      "grad_norm": 0.3156069704147346,
      "learning_rate": 0.00016450839328537172,
      "loss": 1.1263,
      "step": 343
    },
    {
      "epoch": 0.024796367043898218,
      "grad_norm": 0.4396570519375604,
      "learning_rate": 0.00016498800959232615,
      "loss": 1.1696,
      "step": 344
    },
    {
      "epoch": 0.024868449506235134,
      "grad_norm": 0.830464222261114,
      "learning_rate": 0.00016546762589928057,
      "loss": 1.3041,
      "step": 345
    },
    {
      "epoch": 0.024940531968572047,
      "grad_norm": 0.33969587401254353,
      "learning_rate": 0.000165947242206235,
      "loss": 1.1451,
      "step": 346
    },
    {
      "epoch": 0.02501261443090896,
      "grad_norm": 0.6275458128541538,
      "learning_rate": 0.00016642685851318948,
      "loss": 1.1536,
      "step": 347
    },
    {
      "epoch": 0.025084696893245873,
      "grad_norm": 0.8026171609294409,
      "learning_rate": 0.0001669064748201439,
      "loss": 1.1811,
      "step": 348
    },
    {
      "epoch": 0.025156779355582786,
      "grad_norm": 0.5329348703017288,
      "learning_rate": 0.00016738609112709833,
      "loss": 1.1436,
      "step": 349
    },
    {
      "epoch": 0.0252288618179197,
      "grad_norm": 0.28071123929809366,
      "learning_rate": 0.00016786570743405278,
      "loss": 1.1126,
      "step": 350
    },
    {
      "epoch": 0.025300944280256612,
      "grad_norm": 0.7292193956006705,
      "learning_rate": 0.0001683453237410072,
      "loss": 1.2063,
      "step": 351
    },
    {
      "epoch": 0.025373026742593528,
      "grad_norm": 0.5947534182656705,
      "learning_rate": 0.00016882494004796163,
      "loss": 1.2494,
      "step": 352
    },
    {
      "epoch": 0.02544510920493044,
      "grad_norm": 0.4468215313768199,
      "learning_rate": 0.00016930455635491608,
      "loss": 1.1787,
      "step": 353
    },
    {
      "epoch": 0.025517191667267354,
      "grad_norm": 0.3783728094136554,
      "learning_rate": 0.00016978417266187053,
      "loss": 1.162,
      "step": 354
    },
    {
      "epoch": 0.025589274129604267,
      "grad_norm": 0.45891331983843936,
      "learning_rate": 0.00017026378896882496,
      "loss": 1.2129,
      "step": 355
    },
    {
      "epoch": 0.02566135659194118,
      "grad_norm": 0.29919089390469367,
      "learning_rate": 0.00017074340527577938,
      "loss": 1.192,
      "step": 356
    },
    {
      "epoch": 0.025733439054278093,
      "grad_norm": 0.38774785975881343,
      "learning_rate": 0.00017122302158273383,
      "loss": 1.0427,
      "step": 357
    },
    {
      "epoch": 0.02580552151661501,
      "grad_norm": 0.3581518470958333,
      "learning_rate": 0.00017170263788968826,
      "loss": 1.2668,
      "step": 358
    },
    {
      "epoch": 0.025877603978951922,
      "grad_norm": 0.45551514716775227,
      "learning_rate": 0.00017218225419664268,
      "loss": 1.2026,
      "step": 359
    },
    {
      "epoch": 0.025949686441288835,
      "grad_norm": 0.3365961501021499,
      "learning_rate": 0.00017266187050359714,
      "loss": 1.1082,
      "step": 360
    },
    {
      "epoch": 0.026021768903625748,
      "grad_norm": 0.4864390898956754,
      "learning_rate": 0.00017314148681055156,
      "loss": 1.2199,
      "step": 361
    },
    {
      "epoch": 0.02609385136596266,
      "grad_norm": 0.5660757614134259,
      "learning_rate": 0.000173621103117506,
      "loss": 1.2874,
      "step": 362
    },
    {
      "epoch": 0.026165933828299574,
      "grad_norm": 0.939907542526241,
      "learning_rate": 0.00017410071942446044,
      "loss": 1.122,
      "step": 363
    },
    {
      "epoch": 0.026238016290636487,
      "grad_norm": 0.24540492685899246,
      "learning_rate": 0.0001745803357314149,
      "loss": 1.1756,
      "step": 364
    },
    {
      "epoch": 0.026310098752973403,
      "grad_norm": 0.5774760574666385,
      "learning_rate": 0.00017505995203836931,
      "loss": 1.1703,
      "step": 365
    },
    {
      "epoch": 0.026382181215310316,
      "grad_norm": 0.3844508912506733,
      "learning_rate": 0.00017553956834532374,
      "loss": 1.224,
      "step": 366
    },
    {
      "epoch": 0.02645426367764723,
      "grad_norm": 0.39644385215535966,
      "learning_rate": 0.0001760191846522782,
      "loss": 1.2518,
      "step": 367
    },
    {
      "epoch": 0.02652634613998414,
      "grad_norm": 0.27640299509430266,
      "learning_rate": 0.00017649880095923262,
      "loss": 1.1384,
      "step": 368
    },
    {
      "epoch": 0.026598428602321054,
      "grad_norm": 0.43776585728103834,
      "learning_rate": 0.00017697841726618707,
      "loss": 1.2957,
      "step": 369
    },
    {
      "epoch": 0.026670511064657967,
      "grad_norm": 0.2829072500803042,
      "learning_rate": 0.0001774580335731415,
      "loss": 1.0875,
      "step": 370
    },
    {
      "epoch": 0.026742593526994884,
      "grad_norm": 1.0057066163201207,
      "learning_rate": 0.00017793764988009595,
      "loss": 1.1818,
      "step": 371
    },
    {
      "epoch": 0.026814675989331797,
      "grad_norm": 0.26492659646830913,
      "learning_rate": 0.00017841726618705037,
      "loss": 1.1338,
      "step": 372
    },
    {
      "epoch": 0.02688675845166871,
      "grad_norm": 1.3706364381578262,
      "learning_rate": 0.0001788968824940048,
      "loss": 1.1893,
      "step": 373
    },
    {
      "epoch": 0.026958840914005622,
      "grad_norm": 0.20040921216935087,
      "learning_rate": 0.00017937649880095925,
      "loss": 1.1848,
      "step": 374
    },
    {
      "epoch": 0.027030923376342535,
      "grad_norm": 0.2790317256210336,
      "learning_rate": 0.00017985611510791367,
      "loss": 1.1615,
      "step": 375
    },
    {
      "epoch": 0.027103005838679448,
      "grad_norm": 0.32033018766670895,
      "learning_rate": 0.0001803357314148681,
      "loss": 1.1881,
      "step": 376
    },
    {
      "epoch": 0.02717508830101636,
      "grad_norm": 0.5825353265805664,
      "learning_rate": 0.00018081534772182255,
      "loss": 1.2403,
      "step": 377
    },
    {
      "epoch": 0.027247170763353278,
      "grad_norm": 0.5802201283470995,
      "learning_rate": 0.000181294964028777,
      "loss": 1.1418,
      "step": 378
    },
    {
      "epoch": 0.02731925322569019,
      "grad_norm": 0.4110994174909001,
      "learning_rate": 0.00018177458033573143,
      "loss": 1.1824,
      "step": 379
    },
    {
      "epoch": 0.027391335688027103,
      "grad_norm": 0.3046544563361813,
      "learning_rate": 0.00018225419664268585,
      "loss": 1.1781,
      "step": 380
    },
    {
      "epoch": 0.027463418150364016,
      "grad_norm": 1.1965729925914328,
      "learning_rate": 0.0001827338129496403,
      "loss": 1.1622,
      "step": 381
    },
    {
      "epoch": 0.02753550061270093,
      "grad_norm": 0.6440192566097767,
      "learning_rate": 0.00018321342925659473,
      "loss": 1.1364,
      "step": 382
    },
    {
      "epoch": 0.027607583075037842,
      "grad_norm": 0.24644327542523292,
      "learning_rate": 0.00018369304556354915,
      "loss": 1.1702,
      "step": 383
    },
    {
      "epoch": 0.02767966553737476,
      "grad_norm": 0.43789981915372783,
      "learning_rate": 0.0001841726618705036,
      "loss": 1.1739,
      "step": 384
    },
    {
      "epoch": 0.02775174799971167,
      "grad_norm": 0.9272589868732714,
      "learning_rate": 0.00018465227817745806,
      "loss": 1.1897,
      "step": 385
    },
    {
      "epoch": 0.027823830462048584,
      "grad_norm": 0.5288755802793559,
      "learning_rate": 0.00018513189448441248,
      "loss": 1.172,
      "step": 386
    },
    {
      "epoch": 0.027895912924385497,
      "grad_norm": 0.6562387541146064,
      "learning_rate": 0.0001856115107913669,
      "loss": 1.1819,
      "step": 387
    },
    {
      "epoch": 0.02796799538672241,
      "grad_norm": 1.2981780194454344,
      "learning_rate": 0.00018609112709832136,
      "loss": 1.1235,
      "step": 388
    },
    {
      "epoch": 0.028040077849059323,
      "grad_norm": 1.7486912623131965,
      "learning_rate": 0.00018657074340527579,
      "loss": 1.202,
      "step": 389
    },
    {
      "epoch": 0.028112160311396236,
      "grad_norm": 0.387807413561921,
      "learning_rate": 0.0001870503597122302,
      "loss": 1.0869,
      "step": 390
    },
    {
      "epoch": 0.028184242773733152,
      "grad_norm": 0.42538243578168544,
      "learning_rate": 0.00018752997601918466,
      "loss": 1.1215,
      "step": 391
    },
    {
      "epoch": 0.028256325236070065,
      "grad_norm": 0.5940829387834432,
      "learning_rate": 0.00018800959232613911,
      "loss": 1.1248,
      "step": 392
    },
    {
      "epoch": 0.028328407698406978,
      "grad_norm": 0.9299921288748979,
      "learning_rate": 0.00018848920863309354,
      "loss": 1.1391,
      "step": 393
    },
    {
      "epoch": 0.02840049016074389,
      "grad_norm": 1.330086671450345,
      "learning_rate": 0.00018896882494004796,
      "loss": 1.1539,
      "step": 394
    },
    {
      "epoch": 0.028472572623080804,
      "grad_norm": 0.34283601622502613,
      "learning_rate": 0.00018944844124700242,
      "loss": 1.208,
      "step": 395
    },
    {
      "epoch": 0.028544655085417717,
      "grad_norm": 0.6375104008861961,
      "learning_rate": 0.00018992805755395684,
      "loss": 1.0835,
      "step": 396
    },
    {
      "epoch": 0.02861673754775463,
      "grad_norm": 1.8040892645234974,
      "learning_rate": 0.00019040767386091127,
      "loss": 1.1871,
      "step": 397
    },
    {
      "epoch": 0.028688820010091546,
      "grad_norm": 0.31409836880348324,
      "learning_rate": 0.00019088729016786572,
      "loss": 1.2121,
      "step": 398
    },
    {
      "epoch": 0.02876090247242846,
      "grad_norm": 1.274058969707554,
      "learning_rate": 0.00019136690647482017,
      "loss": 1.1288,
      "step": 399
    },
    {
      "epoch": 0.028832984934765372,
      "grad_norm": 0.7301641929551668,
      "learning_rate": 0.0001918465227817746,
      "loss": 1.1279,
      "step": 400
    },
    {
      "epoch": 0.028905067397102285,
      "grad_norm": 0.4776438047449477,
      "learning_rate": 0.00019232613908872902,
      "loss": 1.1513,
      "step": 401
    },
    {
      "epoch": 0.028977149859439198,
      "grad_norm": 1.3433989994951694,
      "learning_rate": 0.00019280575539568347,
      "loss": 1.2085,
      "step": 402
    },
    {
      "epoch": 0.02904923232177611,
      "grad_norm": 1.224971349306326,
      "learning_rate": 0.0001932853717026379,
      "loss": 1.1734,
      "step": 403
    },
    {
      "epoch": 0.029121314784113027,
      "grad_norm": 0.4281489110291011,
      "learning_rate": 0.00019376498800959232,
      "loss": 1.1448,
      "step": 404
    },
    {
      "epoch": 0.02919339724644994,
      "grad_norm": 0.5573032901797548,
      "learning_rate": 0.00019424460431654677,
      "loss": 1.1736,
      "step": 405
    },
    {
      "epoch": 0.029265479708786853,
      "grad_norm": 1.5233465762591087,
      "learning_rate": 0.00019472422062350123,
      "loss": 1.2165,
      "step": 406
    },
    {
      "epoch": 0.029337562171123766,
      "grad_norm": 0.36701813394838456,
      "learning_rate": 0.00019520383693045565,
      "loss": 1.1285,
      "step": 407
    },
    {
      "epoch": 0.02940964463346068,
      "grad_norm": 0.4133672868415404,
      "learning_rate": 0.00019568345323741008,
      "loss": 1.1998,
      "step": 408
    },
    {
      "epoch": 0.02948172709579759,
      "grad_norm": 0.597123804312777,
      "learning_rate": 0.00019616306954436453,
      "loss": 1.3244,
      "step": 409
    },
    {
      "epoch": 0.029553809558134504,
      "grad_norm": 0.21028003823003105,
      "learning_rate": 0.00019664268585131895,
      "loss": 1.1284,
      "step": 410
    },
    {
      "epoch": 0.02962589202047142,
      "grad_norm": 2.083868858632655,
      "learning_rate": 0.00019712230215827338,
      "loss": 1.2097,
      "step": 411
    },
    {
      "epoch": 0.029697974482808333,
      "grad_norm": 0.44069878815074254,
      "learning_rate": 0.00019760191846522783,
      "loss": 1.1596,
      "step": 412
    },
    {
      "epoch": 0.029770056945145246,
      "grad_norm": 1.0155375437912875,
      "learning_rate": 0.00019808153477218226,
      "loss": 1.1272,
      "step": 413
    },
    {
      "epoch": 0.02984213940748216,
      "grad_norm": 0.5995689103134105,
      "learning_rate": 0.0001985611510791367,
      "loss": 1.2039,
      "step": 414
    },
    {
      "epoch": 0.029914221869819072,
      "grad_norm": 1.8402158047146948,
      "learning_rate": 0.00019904076738609113,
      "loss": 1.1951,
      "step": 415
    },
    {
      "epoch": 0.029986304332155985,
      "grad_norm": 0.2578763362576017,
      "learning_rate": 0.00019952038369304558,
      "loss": 1.1619,
      "step": 416
    },
    {
      "epoch": 0.0300583867944929,
      "grad_norm": 0.2572678682470635,
      "learning_rate": 0.0002,
      "loss": 1.1697,
      "step": 417
    },
    {
      "epoch": 0.030130469256829814,
      "grad_norm": 0.3595835961218594,
      "learning_rate": 0.0001999999972745527,
      "loss": 1.1934,
      "step": 418
    },
    {
      "epoch": 0.030202551719166727,
      "grad_norm": 0.9646517970507759,
      "learning_rate": 0.00019999998909821088,
      "loss": 1.2122,
      "step": 419
    },
    {
      "epoch": 0.03027463418150364,
      "grad_norm": 1.416149668574371,
      "learning_rate": 0.0001999999754709751,
      "loss": 1.2447,
      "step": 420
    },
    {
      "epoch": 0.030346716643840553,
      "grad_norm": 1.8247952748200438,
      "learning_rate": 0.000199999956392846,
      "loss": 1.1986,
      "step": 421
    },
    {
      "epoch": 0.030418799106177466,
      "grad_norm": 0.8930420846463194,
      "learning_rate": 0.00019999993186382462,
      "loss": 1.2734,
      "step": 422
    },
    {
      "epoch": 0.03049088156851438,
      "grad_norm": 1.7328591811378202,
      "learning_rate": 0.00019999990188391235,
      "loss": 1.2383,
      "step": 423
    },
    {
      "epoch": 0.030562964030851295,
      "grad_norm": 0.4293590622701056,
      "learning_rate": 0.0001999998664531108,
      "loss": 1.1773,
      "step": 424
    },
    {
      "epoch": 0.030635046493188208,
      "grad_norm": 2.203063963222605,
      "learning_rate": 0.0001999998255714219,
      "loss": 1.1412,
      "step": 425
    },
    {
      "epoch": 0.03070712895552512,
      "grad_norm": 1.9464102296294326,
      "learning_rate": 0.0001999997792388479,
      "loss": 1.1483,
      "step": 426
    },
    {
      "epoch": 0.030779211417862034,
      "grad_norm": 0.4970848352848226,
      "learning_rate": 0.0001999997274553913,
      "loss": 1.1595,
      "step": 427
    },
    {
      "epoch": 0.030851293880198947,
      "grad_norm": 0.4916786087053524,
      "learning_rate": 0.00019999967022105492,
      "loss": 1.1971,
      "step": 428
    },
    {
      "epoch": 0.03092337634253586,
      "grad_norm": 0.3204853109837641,
      "learning_rate": 0.00019999960753584191,
      "loss": 1.1091,
      "step": 429
    },
    {
      "epoch": 0.030995458804872776,
      "grad_norm": 0.6600303502139383,
      "learning_rate": 0.00019999953939975565,
      "loss": 1.1889,
      "step": 430
    },
    {
      "epoch": 0.03106754126720969,
      "grad_norm": 0.26365555496823384,
      "learning_rate": 0.00019999946581279987,
      "loss": 1.1431,
      "step": 431
    },
    {
      "epoch": 0.031139623729546602,
      "grad_norm": 2.295950347074569,
      "learning_rate": 0.00019999938677497857,
      "loss": 1.3063,
      "step": 432
    },
    {
      "epoch": 0.031211706191883515,
      "grad_norm": 0.9157544286524172,
      "learning_rate": 0.0001999993022862961,
      "loss": 1.1064,
      "step": 433
    },
    {
      "epoch": 0.03128378865422043,
      "grad_norm": 0.9156798079324842,
      "learning_rate": 0.00019999921234675703,
      "loss": 1.1851,
      "step": 434
    },
    {
      "epoch": 0.03135587111655734,
      "grad_norm": 0.36498471476722383,
      "learning_rate": 0.00019999911695636627,
      "loss": 1.1134,
      "step": 435
    },
    {
      "epoch": 0.031427953578894254,
      "grad_norm": 0.27428455770215276,
      "learning_rate": 0.000199999016115129,
      "loss": 1.1015,
      "step": 436
    },
    {
      "epoch": 0.031500036041231166,
      "grad_norm": 0.243592483622015,
      "learning_rate": 0.00019999890982305075,
      "loss": 1.1135,
      "step": 437
    },
    {
      "epoch": 0.03157211850356808,
      "grad_norm": 1.0116133001832717,
      "learning_rate": 0.0001999987980801373,
      "loss": 1.1645,
      "step": 438
    },
    {
      "epoch": 0.03164420096590499,
      "grad_norm": 1.9337397847801496,
      "learning_rate": 0.00019999868088639472,
      "loss": 1.215,
      "step": 439
    },
    {
      "epoch": 0.03171628342824191,
      "grad_norm": 3.5045682575951536,
      "learning_rate": 0.00019999855824182948,
      "loss": 1.178,
      "step": 440
    },
    {
      "epoch": 0.031788365890578825,
      "grad_norm": 0.5568977838297189,
      "learning_rate": 0.00019999843014644814,
      "loss": 1.1802,
      "step": 441
    },
    {
      "epoch": 0.03186044835291574,
      "grad_norm": 1.5172977243849366,
      "learning_rate": 0.00019999829660025776,
      "loss": 1.1418,
      "step": 442
    },
    {
      "epoch": 0.03193253081525265,
      "grad_norm": 0.7119565898432121,
      "learning_rate": 0.0001999981576032656,
      "loss": 1.0605,
      "step": 443
    },
    {
      "epoch": 0.032004613277589564,
      "grad_norm": 0.29401999169507215,
      "learning_rate": 0.00019999801315547926,
      "loss": 1.0663,
      "step": 444
    },
    {
      "epoch": 0.03207669573992648,
      "grad_norm": 0.472206602630286,
      "learning_rate": 0.00019999786325690657,
      "loss": 1.2009,
      "step": 445
    },
    {
      "epoch": 0.03214877820226339,
      "grad_norm": 2.5040811751546035,
      "learning_rate": 0.00019999770790755575,
      "loss": 1.1661,
      "step": 446
    },
    {
      "epoch": 0.0322208606646003,
      "grad_norm": 0.45084751213923346,
      "learning_rate": 0.00019999754710743525,
      "loss": 1.181,
      "step": 447
    },
    {
      "epoch": 0.032292943126937215,
      "grad_norm": 1.5919723449452763,
      "learning_rate": 0.0001999973808565538,
      "loss": 1.1119,
      "step": 448
    },
    {
      "epoch": 0.03236502558927413,
      "grad_norm": 1.4129005272033832,
      "learning_rate": 0.0001999972091549205,
      "loss": 1.2013,
      "step": 449
    },
    {
      "epoch": 0.03243710805161104,
      "grad_norm": 1.4110690544515299,
      "learning_rate": 0.0001999970320025447,
      "loss": 1.2156,
      "step": 450
    },
    {
      "epoch": 0.032509190513947954,
      "grad_norm": 0.8803904039168629,
      "learning_rate": 0.00019999684939943607,
      "loss": 1.2455,
      "step": 451
    },
    {
      "epoch": 0.03258127297628487,
      "grad_norm": 0.7385382863422798,
      "learning_rate": 0.00019999666134560453,
      "loss": 1.1956,
      "step": 452
    },
    {
      "epoch": 0.03265335543862179,
      "grad_norm": 0.4915533671456752,
      "learning_rate": 0.00019999646784106035,
      "loss": 1.1718,
      "step": 453
    },
    {
      "epoch": 0.0327254379009587,
      "grad_norm": 0.4127717287672588,
      "learning_rate": 0.00019999626888581414,
      "loss": 1.145,
      "step": 454
    },
    {
      "epoch": 0.03279752036329561,
      "grad_norm": 1.4026247635645552,
      "learning_rate": 0.00019999606447987663,
      "loss": 1.2113,
      "step": 455
    },
    {
      "epoch": 0.032869602825632525,
      "grad_norm": 3.30020546579357,
      "learning_rate": 0.000199995854623259,
      "loss": 1.1964,
      "step": 456
    },
    {
      "epoch": 0.03294168528796944,
      "grad_norm": 0.37696877733430934,
      "learning_rate": 0.0001999956393159727,
      "loss": 1.1227,
      "step": 457
    },
    {
      "epoch": 0.03301376775030635,
      "grad_norm": 0.3593334601557227,
      "learning_rate": 0.00019999541855802953,
      "loss": 1.1627,
      "step": 458
    },
    {
      "epoch": 0.033085850212643264,
      "grad_norm": 1.1594929917388828,
      "learning_rate": 0.00019999519234944143,
      "loss": 1.1858,
      "step": 459
    },
    {
      "epoch": 0.03315793267498018,
      "grad_norm": 1.4265537123143477,
      "learning_rate": 0.00019999496069022076,
      "loss": 1.1164,
      "step": 460
    },
    {
      "epoch": 0.03323001513731709,
      "grad_norm": 1.5664927198049867,
      "learning_rate": 0.00019999472358038015,
      "loss": 1.2526,
      "step": 461
    },
    {
      "epoch": 0.033302097599654,
      "grad_norm": 0.9546528171026845,
      "learning_rate": 0.00019999448101993251,
      "loss": 1.1936,
      "step": 462
    },
    {
      "epoch": 0.033374180061990916,
      "grad_norm": 0.5207315321219828,
      "learning_rate": 0.00019999423300889112,
      "loss": 1.2009,
      "step": 463
    },
    {
      "epoch": 0.03344626252432783,
      "grad_norm": 0.7987442347882956,
      "learning_rate": 0.00019999397954726945,
      "loss": 1.165,
      "step": 464
    },
    {
      "epoch": 0.03351834498666474,
      "grad_norm": 0.6034587614072462,
      "learning_rate": 0.0001999937206350813,
      "loss": 1.2848,
      "step": 465
    },
    {
      "epoch": 0.03359042744900166,
      "grad_norm": 0.7139050705984055,
      "learning_rate": 0.00019999345627234083,
      "loss": 1.2025,
      "step": 466
    },
    {
      "epoch": 0.033662509911338574,
      "grad_norm": 1.3671206316600673,
      "learning_rate": 0.00019999318645906242,
      "loss": 1.1869,
      "step": 467
    },
    {
      "epoch": 0.03373459237367549,
      "grad_norm": 0.6812945867571442,
      "learning_rate": 0.00019999291119526076,
      "loss": 1.3082,
      "step": 468
    },
    {
      "epoch": 0.0338066748360124,
      "grad_norm": 0.45257705286249517,
      "learning_rate": 0.00019999263048095091,
      "loss": 1.168,
      "step": 469
    },
    {
      "epoch": 0.03387875729834931,
      "grad_norm": 0.25060154108035676,
      "learning_rate": 0.00019999234431614812,
      "loss": 1.1406,
      "step": 470
    },
    {
      "epoch": 0.033950839760686226,
      "grad_norm": 0.4501903670377213,
      "learning_rate": 0.000199992052700868,
      "loss": 1.1131,
      "step": 471
    },
    {
      "epoch": 0.03402292222302314,
      "grad_norm": 0.30994859709656314,
      "learning_rate": 0.00019999175563512647,
      "loss": 1.1568,
      "step": 472
    },
    {
      "epoch": 0.03409500468536005,
      "grad_norm": 0.42715790184313385,
      "learning_rate": 0.0001999914531189397,
      "loss": 1.1321,
      "step": 473
    },
    {
      "epoch": 0.034167087147696965,
      "grad_norm": 0.19115665954965075,
      "learning_rate": 0.0001999911451523242,
      "loss": 1.1167,
      "step": 474
    },
    {
      "epoch": 0.03423916961003388,
      "grad_norm": 0.3659659574203984,
      "learning_rate": 0.00019999083173529673,
      "loss": 1.1224,
      "step": 475
    },
    {
      "epoch": 0.03431125207237079,
      "grad_norm": 0.2837137137070841,
      "learning_rate": 0.00019999051286787438,
      "loss": 1.1351,
      "step": 476
    },
    {
      "epoch": 0.0343833345347077,
      "grad_norm": 0.6323765572740576,
      "learning_rate": 0.00019999018855007455,
      "loss": 1.1617,
      "step": 477
    },
    {
      "epoch": 0.034455416997044616,
      "grad_norm": 0.19978572838661704,
      "learning_rate": 0.00019998985878191488,
      "loss": 1.2125,
      "step": 478
    },
    {
      "epoch": 0.03452749945938153,
      "grad_norm": 2.1646136362254853,
      "learning_rate": 0.0001999895235634134,
      "loss": 1.1631,
      "step": 479
    },
    {
      "epoch": 0.03459958192171845,
      "grad_norm": 0.6013692462316027,
      "learning_rate": 0.00019998918289458833,
      "loss": 1.1788,
      "step": 480
    },
    {
      "epoch": 0.03467166438405536,
      "grad_norm": 1.5454402047101043,
      "learning_rate": 0.0001999888367754583,
      "loss": 1.1802,
      "step": 481
    },
    {
      "epoch": 0.034743746846392275,
      "grad_norm": 0.42566854073859356,
      "learning_rate": 0.00019998848520604213,
      "loss": 1.1648,
      "step": 482
    },
    {
      "epoch": 0.03481582930872919,
      "grad_norm": 1.499686084883668,
      "learning_rate": 0.00019998812818635897,
      "loss": 1.1337,
      "step": 483
    },
    {
      "epoch": 0.0348879117710661,
      "grad_norm": 0.29964510592721455,
      "learning_rate": 0.00019998776571642833,
      "loss": 1.1816,
      "step": 484
    },
    {
      "epoch": 0.03495999423340301,
      "grad_norm": 0.6095328614474909,
      "learning_rate": 0.00019998739779626993,
      "loss": 1.1354,
      "step": 485
    },
    {
      "epoch": 0.035032076695739926,
      "grad_norm": 0.35441108322914683,
      "learning_rate": 0.0001999870244259038,
      "loss": 1.1741,
      "step": 486
    },
    {
      "epoch": 0.03510415915807684,
      "grad_norm": 0.3822091049731378,
      "learning_rate": 0.00019998664560535038,
      "loss": 1.1502,
      "step": 487
    },
    {
      "epoch": 0.03517624162041375,
      "grad_norm": 0.28263346869205663,
      "learning_rate": 0.00019998626133463027,
      "loss": 1.22,
      "step": 488
    },
    {
      "epoch": 0.035248324082750665,
      "grad_norm": 0.274289691617912,
      "learning_rate": 0.00019998587161376442,
      "loss": 1.1393,
      "step": 489
    },
    {
      "epoch": 0.03532040654508758,
      "grad_norm": 0.9409891041708651,
      "learning_rate": 0.00019998547644277403,
      "loss": 1.1924,
      "step": 490
    },
    {
      "epoch": 0.03539248900742449,
      "grad_norm": 0.7666279487335712,
      "learning_rate": 0.0001999850758216807,
      "loss": 1.1098,
      "step": 491
    },
    {
      "epoch": 0.035464571469761404,
      "grad_norm": 1.2921868921567605,
      "learning_rate": 0.00019998466975050625,
      "loss": 1.1613,
      "step": 492
    },
    {
      "epoch": 0.035536653932098324,
      "grad_norm": 0.3711535286530877,
      "learning_rate": 0.0001999842582292728,
      "loss": 1.0541,
      "step": 493
    },
    {
      "epoch": 0.035608736394435236,
      "grad_norm": 0.2218175598170067,
      "learning_rate": 0.00019998384125800278,
      "loss": 1.2035,
      "step": 494
    },
    {
      "epoch": 0.03568081885677215,
      "grad_norm": 1.776202346205751,
      "learning_rate": 0.00019998341883671893,
      "loss": 1.1276,
      "step": 495
    },
    {
      "epoch": 0.03575290131910906,
      "grad_norm": 0.6349559690758703,
      "learning_rate": 0.0001999829909654443,
      "loss": 1.2174,
      "step": 496
    },
    {
      "epoch": 0.035824983781445975,
      "grad_norm": 0.3457945451694267,
      "learning_rate": 0.00019998255764420216,
      "loss": 1.0771,
      "step": 497
    },
    {
      "epoch": 0.03589706624378289,
      "grad_norm": 0.49768793451360294,
      "learning_rate": 0.00019998211887301622,
      "loss": 1.2137,
      "step": 498
    },
    {
      "epoch": 0.0359691487061198,
      "grad_norm": 0.21120645759371118,
      "learning_rate": 0.00019998167465191027,
      "loss": 1.1669,
      "step": 499
    },
    {
      "epoch": 0.036041231168456714,
      "grad_norm": 0.9223402515915978,
      "learning_rate": 0.00019998122498090864,
      "loss": 1.2018,
      "step": 500
    },
    {
      "epoch": 0.03611331363079363,
      "grad_norm": 1.4127459706144239,
      "learning_rate": 0.00019998076986003574,
      "loss": 1.2347,
      "step": 501
    },
    {
      "epoch": 0.03618539609313054,
      "grad_norm": 0.4312714005815864,
      "learning_rate": 0.00019998030928931645,
      "loss": 1.2214,
      "step": 502
    },
    {
      "epoch": 0.03625747855546745,
      "grad_norm": 0.3253395874331354,
      "learning_rate": 0.00019997984326877588,
      "loss": 1.0971,
      "step": 503
    },
    {
      "epoch": 0.036329561017804365,
      "grad_norm": 1.7252849910704195,
      "learning_rate": 0.00019997937179843937,
      "loss": 1.206,
      "step": 504
    },
    {
      "epoch": 0.03640164348014128,
      "grad_norm": 1.2508317883358246,
      "learning_rate": 0.00019997889487833267,
      "loss": 1.1787,
      "step": 505
    },
    {
      "epoch": 0.0364737259424782,
      "grad_norm": 1.8342819181481989,
      "learning_rate": 0.00019997841250848176,
      "loss": 1.1216,
      "step": 506
    },
    {
      "epoch": 0.03654580840481511,
      "grad_norm": 0.26219842228676815,
      "learning_rate": 0.00019997792468891295,
      "loss": 1.1264,
      "step": 507
    },
    {
      "epoch": 0.036617890867152024,
      "grad_norm": 0.4224487745006193,
      "learning_rate": 0.0001999774314196528,
      "loss": 1.1532,
      "step": 508
    },
    {
      "epoch": 0.03668997332948894,
      "grad_norm": 0.7092521942064421,
      "learning_rate": 0.00019997693270072824,
      "loss": 1.1385,
      "step": 509
    },
    {
      "epoch": 0.03676205579182585,
      "grad_norm": 0.1942152359793259,
      "learning_rate": 0.00019997642853216641,
      "loss": 1.1861,
      "step": 510
    },
    {
      "epoch": 0.03683413825416276,
      "grad_norm": 0.21137288329492715,
      "learning_rate": 0.00019997591891399478,
      "loss": 1.1445,
      "step": 511
    },
    {
      "epoch": 0.036906220716499676,
      "grad_norm": 0.27594231910713035,
      "learning_rate": 0.0001999754038462412,
      "loss": 1.1393,
      "step": 512
    },
    {
      "epoch": 0.03697830317883659,
      "grad_norm": 0.9515943336980411,
      "learning_rate": 0.00019997488332893371,
      "loss": 1.2003,
      "step": 513
    },
    {
      "epoch": 0.0370503856411735,
      "grad_norm": 0.7242233144831993,
      "learning_rate": 0.00019997435736210064,
      "loss": 1.2513,
      "step": 514
    },
    {
      "epoch": 0.037122468103510414,
      "grad_norm": 0.3613352366512642,
      "learning_rate": 0.00019997382594577073,
      "loss": 1.1918,
      "step": 515
    },
    {
      "epoch": 0.03719455056584733,
      "grad_norm": 1.1694199083453545,
      "learning_rate": 0.0001999732890799729,
      "loss": 1.2131,
      "step": 516
    },
    {
      "epoch": 0.03726663302818424,
      "grad_norm": 0.34369875341614786,
      "learning_rate": 0.0001999727467647364,
      "loss": 1.1683,
      "step": 517
    },
    {
      "epoch": 0.03733871549052115,
      "grad_norm": 1.365052963232434,
      "learning_rate": 0.00019997219900009085,
      "loss": 1.1722,
      "step": 518
    },
    {
      "epoch": 0.03741079795285807,
      "grad_norm": 1.1151641618401353,
      "learning_rate": 0.00019997164578606608,
      "loss": 1.1785,
      "step": 519
    },
    {
      "epoch": 0.037482880415194986,
      "grad_norm": 0.3870675521566558,
      "learning_rate": 0.00019997108712269227,
      "loss": 1.1739,
      "step": 520
    },
    {
      "epoch": 0.0375549628775319,
      "grad_norm": 1.3674241605326096,
      "learning_rate": 0.00019997052300999978,
      "loss": 1.1262,
      "step": 521
    },
    {
      "epoch": 0.03762704533986881,
      "grad_norm": 0.2777628337875105,
      "learning_rate": 0.0001999699534480195,
      "loss": 1.2213,
      "step": 522
    },
    {
      "epoch": 0.037699127802205724,
      "grad_norm": 0.18762491075570456,
      "learning_rate": 0.00019996937843678233,
      "loss": 1.0343,
      "step": 523
    },
    {
      "epoch": 0.03777121026454264,
      "grad_norm": 0.8074765146512499,
      "learning_rate": 0.0001999687979763197,
      "loss": 1.1368,
      "step": 524
    },
    {
      "epoch": 0.03784329272687955,
      "grad_norm": 0.9656720334657308,
      "learning_rate": 0.00019996821206666327,
      "loss": 1.1101,
      "step": 525
    },
    {
      "epoch": 0.03791537518921646,
      "grad_norm": 0.5787668153973672,
      "learning_rate": 0.0001999676207078449,
      "loss": 1.1553,
      "step": 526
    },
    {
      "epoch": 0.037987457651553376,
      "grad_norm": 0.21356976936456534,
      "learning_rate": 0.00019996702389989693,
      "loss": 1.1221,
      "step": 527
    },
    {
      "epoch": 0.03805954011389029,
      "grad_norm": 0.35876161643814586,
      "learning_rate": 0.00019996642164285175,
      "loss": 1.1399,
      "step": 528
    },
    {
      "epoch": 0.0381316225762272,
      "grad_norm": 1.2494378978846847,
      "learning_rate": 0.0001999658139367423,
      "loss": 1.1645,
      "step": 529
    },
    {
      "epoch": 0.038203705038564115,
      "grad_norm": 0.2568645521391823,
      "learning_rate": 0.00019996520078160166,
      "loss": 1.1974,
      "step": 530
    },
    {
      "epoch": 0.03827578750090103,
      "grad_norm": 0.287026685817122,
      "learning_rate": 0.00019996458217746329,
      "loss": 1.1133,
      "step": 531
    },
    {
      "epoch": 0.03834786996323795,
      "grad_norm": 0.3495449078971931,
      "learning_rate": 0.00019996395812436085,
      "loss": 1.1656,
      "step": 532
    },
    {
      "epoch": 0.03841995242557486,
      "grad_norm": 0.68652418220573,
      "learning_rate": 0.0001999633286223284,
      "loss": 1.23,
      "step": 533
    },
    {
      "epoch": 0.03849203488791177,
      "grad_norm": 0.2490615641955298,
      "learning_rate": 0.00019996269367140025,
      "loss": 1.1855,
      "step": 534
    },
    {
      "epoch": 0.038564117350248686,
      "grad_norm": 1.1442289510121004,
      "learning_rate": 0.000199962053271611,
      "loss": 1.1298,
      "step": 535
    },
    {
      "epoch": 0.0386361998125856,
      "grad_norm": 1.0644850062263769,
      "learning_rate": 0.00019996140742299553,
      "loss": 1.1197,
      "step": 536
    },
    {
      "epoch": 0.03870828227492251,
      "grad_norm": 0.23017478541129271,
      "learning_rate": 0.00019996075612558915,
      "loss": 1.0625,
      "step": 537
    },
    {
      "epoch": 0.038780364737259425,
      "grad_norm": 0.20283913537892817,
      "learning_rate": 0.00019996009937942724,
      "loss": 1.1579,
      "step": 538
    },
    {
      "epoch": 0.03885244719959634,
      "grad_norm": 0.40531994587680803,
      "learning_rate": 0.00019995943718454563,
      "loss": 1.1657,
      "step": 539
    },
    {
      "epoch": 0.03892452966193325,
      "grad_norm": 0.2656233199237605,
      "learning_rate": 0.00019995876954098044,
      "loss": 1.0932,
      "step": 540
    },
    {
      "epoch": 0.038996612124270164,
      "grad_norm": 0.22091829632795423,
      "learning_rate": 0.00019995809644876805,
      "loss": 1.1105,
      "step": 541
    },
    {
      "epoch": 0.039068694586607076,
      "grad_norm": 1.6522440447142888,
      "learning_rate": 0.00019995741790794518,
      "loss": 1.193,
      "step": 542
    },
    {
      "epoch": 0.03914077704894399,
      "grad_norm": 0.39309642134212835,
      "learning_rate": 0.00019995673391854876,
      "loss": 1.0846,
      "step": 543
    },
    {
      "epoch": 0.0392128595112809,
      "grad_norm": 0.15910260210581903,
      "learning_rate": 0.00019995604448061612,
      "loss": 1.1585,
      "step": 544
    },
    {
      "epoch": 0.03928494197361782,
      "grad_norm": 0.47841661621128306,
      "learning_rate": 0.00019995534959418482,
      "loss": 1.2369,
      "step": 545
    },
    {
      "epoch": 0.039357024435954735,
      "grad_norm": 0.22324757994945055,
      "learning_rate": 0.00019995464925929275,
      "loss": 1.2079,
      "step": 546
    },
    {
      "epoch": 0.03942910689829165,
      "grad_norm": 0.20716247117663972,
      "learning_rate": 0.00019995394347597804,
      "loss": 1.1392,
      "step": 547
    },
    {
      "epoch": 0.03950118936062856,
      "grad_norm": 0.34837935153738275,
      "learning_rate": 0.00019995323224427923,
      "loss": 1.1612,
      "step": 548
    },
    {
      "epoch": 0.039573271822965474,
      "grad_norm": 0.4396094029822607,
      "learning_rate": 0.00019995251556423503,
      "loss": 1.1779,
      "step": 549
    },
    {
      "epoch": 0.03964535428530239,
      "grad_norm": 0.7073519863788625,
      "learning_rate": 0.00019995179343588454,
      "loss": 1.1148,
      "step": 550
    },
    {
      "epoch": 0.0397174367476393,
      "grad_norm": 0.26741246398872176,
      "learning_rate": 0.00019995106585926715,
      "loss": 1.115,
      "step": 551
    },
    {
      "epoch": 0.03978951920997621,
      "grad_norm": 1.2520824520950564,
      "learning_rate": 0.00019995033283442247,
      "loss": 1.0899,
      "step": 552
    },
    {
      "epoch": 0.039861601672313125,
      "grad_norm": 0.4145372348538698,
      "learning_rate": 0.00019994959436139045,
      "loss": 1.139,
      "step": 553
    },
    {
      "epoch": 0.03993368413465004,
      "grad_norm": 0.25357012850873256,
      "learning_rate": 0.00019994885044021134,
      "loss": 1.1366,
      "step": 554
    },
    {
      "epoch": 0.04000576659698695,
      "grad_norm": 0.3693177915531279,
      "learning_rate": 0.00019994810107092575,
      "loss": 1.0573,
      "step": 555
    },
    {
      "epoch": 0.040077849059323864,
      "grad_norm": 0.27249326008163716,
      "learning_rate": 0.0001999473462535745,
      "loss": 1.1303,
      "step": 556
    },
    {
      "epoch": 0.04014993152166078,
      "grad_norm": 1.4005339450377399,
      "learning_rate": 0.0001999465859881987,
      "loss": 1.1726,
      "step": 557
    },
    {
      "epoch": 0.0402220139839977,
      "grad_norm": 1.0436136068055484,
      "learning_rate": 0.0001999458202748398,
      "loss": 1.1492,
      "step": 558
    },
    {
      "epoch": 0.04029409644633461,
      "grad_norm": 0.18706359477897172,
      "learning_rate": 0.0001999450491135396,
      "loss": 1.161,
      "step": 559
    },
    {
      "epoch": 0.04036617890867152,
      "grad_norm": 1.8698532403763648,
      "learning_rate": 0.00019994427250434005,
      "loss": 1.1277,
      "step": 560
    },
    {
      "epoch": 0.040438261371008435,
      "grad_norm": 0.6190609154780742,
      "learning_rate": 0.00019994349044728354,
      "loss": 1.1575,
      "step": 561
    },
    {
      "epoch": 0.04051034383334535,
      "grad_norm": 1.211744152785148,
      "learning_rate": 0.00019994270294241266,
      "loss": 1.1493,
      "step": 562
    },
    {
      "epoch": 0.04058242629568226,
      "grad_norm": 0.3078700565030194,
      "learning_rate": 0.0001999419099897704,
      "loss": 1.1216,
      "step": 563
    },
    {
      "epoch": 0.040654508758019174,
      "grad_norm": 3.6474904660508467,
      "learning_rate": 0.00019994111158939989,
      "loss": 1.1942,
      "step": 564
    },
    {
      "epoch": 0.04072659122035609,
      "grad_norm": 1.5822221374607783,
      "learning_rate": 0.00019994030774134474,
      "loss": 1.2067,
      "step": 565
    },
    {
      "epoch": 0.040798673682693,
      "grad_norm": 1.655269166536514,
      "learning_rate": 0.00019993949844564869,
      "loss": 1.1647,
      "step": 566
    },
    {
      "epoch": 0.04087075614502991,
      "grad_norm": 0.41357869443728756,
      "learning_rate": 0.00019993868370235592,
      "loss": 1.1952,
      "step": 567
    },
    {
      "epoch": 0.040942838607366826,
      "grad_norm": 0.180003429327125,
      "learning_rate": 0.00019993786351151077,
      "loss": 1.1379,
      "step": 568
    },
    {
      "epoch": 0.04101492106970374,
      "grad_norm": 0.16697948489068654,
      "learning_rate": 0.00019993703787315803,
      "loss": 1.1818,
      "step": 569
    },
    {
      "epoch": 0.04108700353204065,
      "grad_norm": 0.5616976379863162,
      "learning_rate": 0.00019993620678734265,
      "loss": 1.1715,
      "step": 570
    },
    {
      "epoch": 0.04115908599437757,
      "grad_norm": 1.16521212985531,
      "learning_rate": 0.00019993537025410997,
      "loss": 1.2251,
      "step": 571
    },
    {
      "epoch": 0.041231168456714484,
      "grad_norm": 0.4092406533159282,
      "learning_rate": 0.00019993452827350553,
      "loss": 1.209,
      "step": 572
    },
    {
      "epoch": 0.0413032509190514,
      "grad_norm": 0.40256770137747827,
      "learning_rate": 0.00019993368084557527,
      "loss": 1.0635,
      "step": 573
    },
    {
      "epoch": 0.04137533338138831,
      "grad_norm": 0.9349259536657936,
      "learning_rate": 0.0001999328279703654,
      "loss": 1.1277,
      "step": 574
    },
    {
      "epoch": 0.04144741584372522,
      "grad_norm": 3.2683845937832143,
      "learning_rate": 0.00019993196964792235,
      "loss": 1.1729,
      "step": 575
    },
    {
      "epoch": 0.041519498306062136,
      "grad_norm": 0.6392152845676786,
      "learning_rate": 0.00019993110587829293,
      "loss": 1.0674,
      "step": 576
    },
    {
      "epoch": 0.04159158076839905,
      "grad_norm": 0.26654834877980793,
      "learning_rate": 0.00019993023666152425,
      "loss": 1.2167,
      "step": 577
    },
    {
      "epoch": 0.04166366323073596,
      "grad_norm": 0.9164407253709262,
      "learning_rate": 0.00019992936199766365,
      "loss": 1.1597,
      "step": 578
    },
    {
      "epoch": 0.041735745693072875,
      "grad_norm": 0.26759652458384237,
      "learning_rate": 0.00019992848188675884,
      "loss": 1.1475,
      "step": 579
    },
    {
      "epoch": 0.04180782815540979,
      "grad_norm": 2.094740576960766,
      "learning_rate": 0.0001999275963288578,
      "loss": 1.1772,
      "step": 580
    },
    {
      "epoch": 0.0418799106177467,
      "grad_norm": 0.8230535069511752,
      "learning_rate": 0.00019992670532400874,
      "loss": 1.3185,
      "step": 581
    },
    {
      "epoch": 0.04195199308008361,
      "grad_norm": 0.21542129650852254,
      "learning_rate": 0.0001999258088722603,
      "loss": 1.161,
      "step": 582
    },
    {
      "epoch": 0.042024075542420526,
      "grad_norm": 0.8111108400025141,
      "learning_rate": 0.00019992490697366127,
      "loss": 1.2172,
      "step": 583
    },
    {
      "epoch": 0.04209615800475744,
      "grad_norm": 0.8915131893758604,
      "learning_rate": 0.0001999239996282609,
      "loss": 1.2835,
      "step": 584
    },
    {
      "epoch": 0.04216824046709436,
      "grad_norm": 1.6794757285282855,
      "learning_rate": 0.0001999230868361086,
      "loss": 1.2276,
      "step": 585
    },
    {
      "epoch": 0.04224032292943127,
      "grad_norm": 0.36709235116315997,
      "learning_rate": 0.00019992216859725413,
      "loss": 1.2442,
      "step": 586
    },
    {
      "epoch": 0.042312405391768185,
      "grad_norm": 0.8127818387477795,
      "learning_rate": 0.00019992124491174758,
      "loss": 1.0819,
      "step": 587
    },
    {
      "epoch": 0.0423844878541051,
      "grad_norm": 2.038709747307564,
      "learning_rate": 0.00019992031577963922,
      "loss": 1.2427,
      "step": 588
    },
    {
      "epoch": 0.04245657031644201,
      "grad_norm": 0.1715150790890134,
      "learning_rate": 0.00019991938120097973,
      "loss": 1.1869,
      "step": 589
    },
    {
      "epoch": 0.04252865277877892,
      "grad_norm": 2.6886523328284873,
      "learning_rate": 0.00019991844117582007,
      "loss": 1.2052,
      "step": 590
    },
    {
      "epoch": 0.042600735241115836,
      "grad_norm": 2.1561520237304395,
      "learning_rate": 0.00019991749570421146,
      "loss": 1.2211,
      "step": 591
    },
    {
      "epoch": 0.04267281770345275,
      "grad_norm": 0.4779106902932627,
      "learning_rate": 0.00019991654478620547,
      "loss": 1.0616,
      "step": 592
    },
    {
      "epoch": 0.04274490016578966,
      "grad_norm": 0.48622187761293933,
      "learning_rate": 0.00019991558842185388,
      "loss": 1.1824,
      "step": 593
    },
    {
      "epoch": 0.042816982628126575,
      "grad_norm": 0.9985166275776874,
      "learning_rate": 0.00019991462661120887,
      "loss": 1.1643,
      "step": 594
    },
    {
      "epoch": 0.04288906509046349,
      "grad_norm": 0.5422837706055574,
      "learning_rate": 0.00019991365935432284,
      "loss": 1.206,
      "step": 595
    },
    {
      "epoch": 0.0429611475528004,
      "grad_norm": 0.1922701761272907,
      "learning_rate": 0.00019991268665124853,
      "loss": 1.1527,
      "step": 596
    },
    {
      "epoch": 0.043033230015137314,
      "grad_norm": 0.28146133649035787,
      "learning_rate": 0.00019991170850203896,
      "loss": 1.1971,
      "step": 597
    },
    {
      "epoch": 0.043105312477474234,
      "grad_norm": 0.9317520154679493,
      "learning_rate": 0.00019991072490674742,
      "loss": 1.1878,
      "step": 598
    },
    {
      "epoch": 0.043177394939811146,
      "grad_norm": 0.48320482102957785,
      "learning_rate": 0.00019990973586542755,
      "loss": 1.1483,
      "step": 599
    },
    {
      "epoch": 0.04324947740214806,
      "grad_norm": 0.29315876730765245,
      "learning_rate": 0.00019990874137813328,
      "loss": 1.2276,
      "step": 600
    },
    {
      "epoch": 0.04332155986448497,
      "grad_norm": 0.34757398303180803,
      "learning_rate": 0.00019990774144491876,
      "loss": 1.2544,
      "step": 601
    },
    {
      "epoch": 0.043393642326821885,
      "grad_norm": 0.5783617956632899,
      "learning_rate": 0.00019990673606583856,
      "loss": 1.1429,
      "step": 602
    },
    {
      "epoch": 0.0434657247891588,
      "grad_norm": 1.2575829634454823,
      "learning_rate": 0.00019990572524094743,
      "loss": 1.0975,
      "step": 603
    },
    {
      "epoch": 0.04353780725149571,
      "grad_norm": 0.5542510695472169,
      "learning_rate": 0.0001999047089703005,
      "loss": 1.2014,
      "step": 604
    },
    {
      "epoch": 0.043609889713832624,
      "grad_norm": 0.444081449262812,
      "learning_rate": 0.00019990368725395318,
      "loss": 1.152,
      "step": 605
    },
    {
      "epoch": 0.04368197217616954,
      "grad_norm": 1.4765479419797825,
      "learning_rate": 0.00019990266009196114,
      "loss": 1.0755,
      "step": 606
    },
    {
      "epoch": 0.04375405463850645,
      "grad_norm": 0.5260997108126598,
      "learning_rate": 0.00019990162748438034,
      "loss": 1.1112,
      "step": 607
    },
    {
      "epoch": 0.04382613710084336,
      "grad_norm": 0.3900005423575157,
      "learning_rate": 0.0001999005894312671,
      "loss": 1.2299,
      "step": 608
    },
    {
      "epoch": 0.043898219563180275,
      "grad_norm": 0.23508354168966802,
      "learning_rate": 0.00019989954593267804,
      "loss": 1.1331,
      "step": 609
    },
    {
      "epoch": 0.04397030202551719,
      "grad_norm": 0.765471843795879,
      "learning_rate": 0.00019989849698866995,
      "loss": 1.0779,
      "step": 610
    },
    {
      "epoch": 0.04404238448785411,
      "grad_norm": 0.22444684468514983,
      "learning_rate": 0.0001998974425993001,
      "loss": 1.1093,
      "step": 611
    },
    {
      "epoch": 0.04411446695019102,
      "grad_norm": 0.20423131581943432,
      "learning_rate": 0.0001998963827646259,
      "loss": 1.1173,
      "step": 612
    },
    {
      "epoch": 0.044186549412527934,
      "grad_norm": 0.2540325511176254,
      "learning_rate": 0.00019989531748470515,
      "loss": 1.1672,
      "step": 613
    },
    {
      "epoch": 0.04425863187486485,
      "grad_norm": 0.42330236789765646,
      "learning_rate": 0.00019989424675959588,
      "loss": 1.1291,
      "step": 614
    },
    {
      "epoch": 0.04433071433720176,
      "grad_norm": 0.4754168900758251,
      "learning_rate": 0.00019989317058935648,
      "loss": 1.1496,
      "step": 615
    },
    {
      "epoch": 0.04440279679953867,
      "grad_norm": 1.3889937398199905,
      "learning_rate": 0.00019989208897404565,
      "loss": 1.2014,
      "step": 616
    },
    {
      "epoch": 0.044474879261875586,
      "grad_norm": 0.9387386708490167,
      "learning_rate": 0.0001998910019137223,
      "loss": 1.1703,
      "step": 617
    },
    {
      "epoch": 0.0445469617242125,
      "grad_norm": 0.19831112498658895,
      "learning_rate": 0.00019988990940844568,
      "loss": 1.0839,
      "step": 618
    },
    {
      "epoch": 0.04461904418654941,
      "grad_norm": 0.18083998228417364,
      "learning_rate": 0.00019988881145827536,
      "loss": 1.1285,
      "step": 619
    },
    {
      "epoch": 0.044691126648886324,
      "grad_norm": 0.2490397069645933,
      "learning_rate": 0.0001998877080632712,
      "loss": 1.0886,
      "step": 620
    },
    {
      "epoch": 0.04476320911122324,
      "grad_norm": 0.1794455507433621,
      "learning_rate": 0.0001998865992234933,
      "loss": 1.1705,
      "step": 621
    },
    {
      "epoch": 0.04483529157356015,
      "grad_norm": 0.6047213912964543,
      "learning_rate": 0.00019988548493900216,
      "loss": 1.1188,
      "step": 622
    },
    {
      "epoch": 0.04490737403589706,
      "grad_norm": 0.18976442520045672,
      "learning_rate": 0.0001998843652098585,
      "loss": 1.1367,
      "step": 623
    },
    {
      "epoch": 0.04497945649823398,
      "grad_norm": 0.27411278623592766,
      "learning_rate": 0.0001998832400361233,
      "loss": 1.0498,
      "step": 624
    },
    {
      "epoch": 0.045051538960570896,
      "grad_norm": 0.40388652578945017,
      "learning_rate": 0.00019988210941785796,
      "loss": 1.1099,
      "step": 625
    },
    {
      "epoch": 0.04512362142290781,
      "grad_norm": 0.7890063162121763,
      "learning_rate": 0.0001998809733551241,
      "loss": 1.0691,
      "step": 626
    },
    {
      "epoch": 0.04519570388524472,
      "grad_norm": 1.640346487263937,
      "learning_rate": 0.00019987983184798362,
      "loss": 1.1195,
      "step": 627
    },
    {
      "epoch": 0.045267786347581634,
      "grad_norm": 0.20877723464196993,
      "learning_rate": 0.00019987868489649874,
      "loss": 1.1489,
      "step": 628
    },
    {
      "epoch": 0.04533986880991855,
      "grad_norm": 0.4005833677383866,
      "learning_rate": 0.000199877532500732,
      "loss": 1.1771,
      "step": 629
    },
    {
      "epoch": 0.04541195127225546,
      "grad_norm": 0.2589916959540075,
      "learning_rate": 0.00019987637466074626,
      "loss": 1.169,
      "step": 630
    },
    {
      "epoch": 0.04548403373459237,
      "grad_norm": 1.0414007745524168,
      "learning_rate": 0.00019987521137660453,
      "loss": 1.0746,
      "step": 631
    },
    {
      "epoch": 0.045556116196929286,
      "grad_norm": 0.17938515818427647,
      "learning_rate": 0.00019987404264837025,
      "loss": 1.1736,
      "step": 632
    },
    {
      "epoch": 0.0456281986592662,
      "grad_norm": 0.30898045031250215,
      "learning_rate": 0.00019987286847610716,
      "loss": 1.0768,
      "step": 633
    },
    {
      "epoch": 0.04570028112160311,
      "grad_norm": 0.15886790072473095,
      "learning_rate": 0.00019987168885987928,
      "loss": 1.1746,
      "step": 634
    },
    {
      "epoch": 0.045772363583940025,
      "grad_norm": 0.20590184498875386,
      "learning_rate": 0.00019987050379975085,
      "loss": 1.1328,
      "step": 635
    },
    {
      "epoch": 0.04584444604627694,
      "grad_norm": 0.15875730842770736,
      "learning_rate": 0.0001998693132957865,
      "loss": 1.117,
      "step": 636
    },
    {
      "epoch": 0.04591652850861386,
      "grad_norm": 1.854945140478252,
      "learning_rate": 0.00019986811734805116,
      "loss": 1.1696,
      "step": 637
    },
    {
      "epoch": 0.04598861097095077,
      "grad_norm": 0.3726396149682728,
      "learning_rate": 0.00019986691595660994,
      "loss": 1.171,
      "step": 638
    },
    {
      "epoch": 0.04606069343328768,
      "grad_norm": 0.15818019450541182,
      "learning_rate": 0.00019986570912152837,
      "loss": 1.0507,
      "step": 639
    },
    {
      "epoch": 0.046132775895624596,
      "grad_norm": 0.2586362420758278,
      "learning_rate": 0.00019986449684287222,
      "loss": 1.155,
      "step": 640
    },
    {
      "epoch": 0.04620485835796151,
      "grad_norm": 1.0534986457309345,
      "learning_rate": 0.0001998632791207076,
      "loss": 1.0911,
      "step": 641
    },
    {
      "epoch": 0.04627694082029842,
      "grad_norm": 0.2076275587449345,
      "learning_rate": 0.00019986205595510086,
      "loss": 1.107,
      "step": 642
    },
    {
      "epoch": 0.046349023282635335,
      "grad_norm": 0.23277170556504834,
      "learning_rate": 0.00019986082734611867,
      "loss": 1.0915,
      "step": 643
    },
    {
      "epoch": 0.04642110574497225,
      "grad_norm": 0.253406438222046,
      "learning_rate": 0.000199859593293828,
      "loss": 1.2137,
      "step": 644
    },
    {
      "epoch": 0.04649318820730916,
      "grad_norm": 0.3371528070543081,
      "learning_rate": 0.00019985835379829616,
      "loss": 1.1645,
      "step": 645
    },
    {
      "epoch": 0.046565270669646074,
      "grad_norm": 0.17477369859592184,
      "learning_rate": 0.00019985710885959065,
      "loss": 1.2001,
      "step": 646
    },
    {
      "epoch": 0.046637353131982986,
      "grad_norm": 0.2755097835210729,
      "learning_rate": 0.0001998558584777794,
      "loss": 1.1249,
      "step": 647
    },
    {
      "epoch": 0.0467094355943199,
      "grad_norm": 0.16966536152224698,
      "learning_rate": 0.00019985460265293048,
      "loss": 1.1271,
      "step": 648
    },
    {
      "epoch": 0.04678151805665681,
      "grad_norm": 0.488144395999296,
      "learning_rate": 0.00019985334138511237,
      "loss": 1.1306,
      "step": 649
    },
    {
      "epoch": 0.04685360051899373,
      "grad_norm": 0.20099740323326212,
      "learning_rate": 0.0001998520746743939,
      "loss": 1.151,
      "step": 650
    },
    {
      "epoch": 0.046925682981330645,
      "grad_norm": 0.18972263070089262,
      "learning_rate": 0.00019985080252084406,
      "loss": 1.0541,
      "step": 651
    },
    {
      "epoch": 0.04699776544366756,
      "grad_norm": 0.1408659190550918,
      "learning_rate": 0.00019984952492453216,
      "loss": 1.1355,
      "step": 652
    },
    {
      "epoch": 0.04706984790600447,
      "grad_norm": 0.16428706561589346,
      "learning_rate": 0.00019984824188552786,
      "loss": 1.1337,
      "step": 653
    },
    {
      "epoch": 0.047141930368341384,
      "grad_norm": 0.42751615697774875,
      "learning_rate": 0.00019984695340390113,
      "loss": 1.1022,
      "step": 654
    },
    {
      "epoch": 0.0472140128306783,
      "grad_norm": 0.20284695253487647,
      "learning_rate": 0.00019984565947972217,
      "loss": 1.1732,
      "step": 655
    },
    {
      "epoch": 0.04728609529301521,
      "grad_norm": 0.14181648803059016,
      "learning_rate": 0.00019984436011306157,
      "loss": 1.1039,
      "step": 656
    },
    {
      "epoch": 0.04735817775535212,
      "grad_norm": 0.2868471154838052,
      "learning_rate": 0.00019984305530399007,
      "loss": 1.05,
      "step": 657
    },
    {
      "epoch": 0.047430260217689035,
      "grad_norm": 0.9961338579104061,
      "learning_rate": 0.00019984174505257882,
      "loss": 1.1546,
      "step": 658
    },
    {
      "epoch": 0.04750234268002595,
      "grad_norm": 0.29430961873389544,
      "learning_rate": 0.00019984042935889928,
      "loss": 1.1297,
      "step": 659
    },
    {
      "epoch": 0.04757442514236286,
      "grad_norm": 0.24153588724561273,
      "learning_rate": 0.00019983910822302315,
      "loss": 1.1446,
      "step": 660
    },
    {
      "epoch": 0.047646507604699774,
      "grad_norm": 0.8336139651046132,
      "learning_rate": 0.0001998377816450224,
      "loss": 1.2247,
      "step": 661
    },
    {
      "epoch": 0.04771859006703669,
      "grad_norm": 0.1644468000234561,
      "learning_rate": 0.00019983644962496942,
      "loss": 1.1928,
      "step": 662
    },
    {
      "epoch": 0.04779067252937361,
      "grad_norm": 0.22781100897837564,
      "learning_rate": 0.00019983511216293675,
      "loss": 1.0603,
      "step": 663
    },
    {
      "epoch": 0.04786275499171052,
      "grad_norm": 0.18118464087063654,
      "learning_rate": 0.00019983376925899733,
      "loss": 1.1038,
      "step": 664
    },
    {
      "epoch": 0.04793483745404743,
      "grad_norm": 1.5079393349658405,
      "learning_rate": 0.00019983242091322433,
      "loss": 1.1209,
      "step": 665
    },
    {
      "epoch": 0.048006919916384345,
      "grad_norm": 0.14435801817807517,
      "learning_rate": 0.00019983106712569127,
      "loss": 1.0517,
      "step": 666
    },
    {
      "epoch": 0.04807900237872126,
      "grad_norm": 0.1986282659998528,
      "learning_rate": 0.00019982970789647194,
      "loss": 1.1268,
      "step": 667
    },
    {
      "epoch": 0.04815108484105817,
      "grad_norm": 0.1737479745029533,
      "learning_rate": 0.0001998283432256404,
      "loss": 1.0881,
      "step": 668
    },
    {
      "epoch": 0.048223167303395084,
      "grad_norm": 0.7584655994073549,
      "learning_rate": 0.0001998269731132711,
      "loss": 1.1426,
      "step": 669
    },
    {
      "epoch": 0.048295249765732,
      "grad_norm": 0.2146831578365741,
      "learning_rate": 0.0001998255975594387,
      "loss": 1.1209,
      "step": 670
    },
    {
      "epoch": 0.04836733222806891,
      "grad_norm": 0.3127187208752,
      "learning_rate": 0.0001998242165642181,
      "loss": 1.1498,
      "step": 671
    },
    {
      "epoch": 0.04843941469040582,
      "grad_norm": 0.1695680546457042,
      "learning_rate": 0.00019982283012768472,
      "loss": 1.1488,
      "step": 672
    },
    {
      "epoch": 0.048511497152742736,
      "grad_norm": 0.3270808759326753,
      "learning_rate": 0.00019982143824991402,
      "loss": 1.0942,
      "step": 673
    },
    {
      "epoch": 0.04858357961507965,
      "grad_norm": 0.14693256579513617,
      "learning_rate": 0.00019982004093098193,
      "loss": 1.185,
      "step": 674
    },
    {
      "epoch": 0.04865566207741656,
      "grad_norm": 1.7536297824955172,
      "learning_rate": 0.00019981863817096458,
      "loss": 1.1446,
      "step": 675
    },
    {
      "epoch": 0.04872774453975348,
      "grad_norm": 0.25043189427046697,
      "learning_rate": 0.00019981722996993844,
      "loss": 1.1639,
      "step": 676
    },
    {
      "epoch": 0.048799827002090394,
      "grad_norm": 1.2292696121529088,
      "learning_rate": 0.00019981581632798028,
      "loss": 1.2144,
      "step": 677
    },
    {
      "epoch": 0.04887190946442731,
      "grad_norm": 0.17406066923172808,
      "learning_rate": 0.00019981439724516716,
      "loss": 1.1373,
      "step": 678
    },
    {
      "epoch": 0.04894399192676422,
      "grad_norm": 0.19884729869114212,
      "learning_rate": 0.00019981297272157639,
      "loss": 1.1915,
      "step": 679
    },
    {
      "epoch": 0.04901607438910113,
      "grad_norm": 0.6894550413670252,
      "learning_rate": 0.0001998115427572857,
      "loss": 1.1773,
      "step": 680
    },
    {
      "epoch": 0.049088156851438046,
      "grad_norm": 0.24170612031862063,
      "learning_rate": 0.00019981010735237295,
      "loss": 1.1403,
      "step": 681
    },
    {
      "epoch": 0.04916023931377496,
      "grad_norm": 0.26146775032404496,
      "learning_rate": 0.00019980866650691647,
      "loss": 1.1595,
      "step": 682
    },
    {
      "epoch": 0.04923232177611187,
      "grad_norm": 0.32976777484280667,
      "learning_rate": 0.0001998072202209947,
      "loss": 1.1029,
      "step": 683
    },
    {
      "epoch": 0.049304404238448785,
      "grad_norm": 0.33040613963495064,
      "learning_rate": 0.00019980576849468656,
      "loss": 1.1545,
      "step": 684
    },
    {
      "epoch": 0.0493764867007857,
      "grad_norm": 0.500227731635009,
      "learning_rate": 0.00019980431132807117,
      "loss": 1.2107,
      "step": 685
    },
    {
      "epoch": 0.04944856916312261,
      "grad_norm": 0.15695566780518844,
      "learning_rate": 0.00019980284872122788,
      "loss": 1.0855,
      "step": 686
    },
    {
      "epoch": 0.04952065162545952,
      "grad_norm": 1.2785103635878379,
      "learning_rate": 0.0001998013806742365,
      "loss": 1.1715,
      "step": 687
    },
    {
      "epoch": 0.049592734087796436,
      "grad_norm": 0.25653615057274637,
      "learning_rate": 0.00019979990718717704,
      "loss": 1.1411,
      "step": 688
    },
    {
      "epoch": 0.04966481655013335,
      "grad_norm": 0.27567127267054703,
      "learning_rate": 0.00019979842826012978,
      "loss": 1.1734,
      "step": 689
    },
    {
      "epoch": 0.04973689901247027,
      "grad_norm": 0.5380555028502348,
      "learning_rate": 0.00019979694389317537,
      "loss": 1.112,
      "step": 690
    },
    {
      "epoch": 0.04980898147480718,
      "grad_norm": 0.2996132848525292,
      "learning_rate": 0.00019979545408639471,
      "loss": 1.2124,
      "step": 691
    },
    {
      "epoch": 0.049881063937144095,
      "grad_norm": 1.7322539544873496,
      "learning_rate": 0.000199793958839869,
      "loss": 1.1411,
      "step": 692
    },
    {
      "epoch": 0.04995314639948101,
      "grad_norm": 0.9952038532256956,
      "learning_rate": 0.00019979245815367978,
      "loss": 1.225,
      "step": 693
    },
    {
      "epoch": 0.05002522886181792,
      "grad_norm": 0.14705646279820045,
      "learning_rate": 0.00019979095202790877,
      "loss": 1.071,
      "step": 694
    },
    {
      "epoch": 0.05009731132415483,
      "grad_norm": 0.2176620871917936,
      "learning_rate": 0.00019978944046263815,
      "loss": 1.1448,
      "step": 695
    },
    {
      "epoch": 0.050169393786491746,
      "grad_norm": 0.33531090933868335,
      "learning_rate": 0.0001997879234579503,
      "loss": 1.1375,
      "step": 696
    },
    {
      "epoch": 0.05024147624882866,
      "grad_norm": 0.24204689120296427,
      "learning_rate": 0.00019978640101392785,
      "loss": 1.1038,
      "step": 697
    },
    {
      "epoch": 0.05031355871116557,
      "grad_norm": 1.2595686584864951,
      "learning_rate": 0.00019978487313065385,
      "loss": 1.2278,
      "step": 698
    },
    {
      "epoch": 0.050385641173502485,
      "grad_norm": 1.3092524013659037,
      "learning_rate": 0.00019978333980821158,
      "loss": 1.1061,
      "step": 699
    },
    {
      "epoch": 0.0504577236358394,
      "grad_norm": 0.37184730926136544,
      "learning_rate": 0.00019978180104668458,
      "loss": 1.2035,
      "step": 700
    },
    {
      "epoch": 0.05052980609817631,
      "grad_norm": 1.370768210766976,
      "learning_rate": 0.00019978025684615674,
      "loss": 1.1728,
      "step": 701
    },
    {
      "epoch": 0.050601888560513224,
      "grad_norm": 0.1584479381157944,
      "learning_rate": 0.00019977870720671228,
      "loss": 1.1692,
      "step": 702
    },
    {
      "epoch": 0.050673971022850144,
      "grad_norm": 0.2580634395639809,
      "learning_rate": 0.0001997771521284356,
      "loss": 1.2031,
      "step": 703
    },
    {
      "epoch": 0.050746053485187056,
      "grad_norm": 0.21381612937109706,
      "learning_rate": 0.00019977559161141152,
      "loss": 1.1881,
      "step": 704
    },
    {
      "epoch": 0.05081813594752397,
      "grad_norm": 0.2859421800977926,
      "learning_rate": 0.00019977402565572505,
      "loss": 1.1166,
      "step": 705
    },
    {
      "epoch": 0.05089021840986088,
      "grad_norm": 0.43534469556785815,
      "learning_rate": 0.00019977245426146161,
      "loss": 1.2108,
      "step": 706
    },
    {
      "epoch": 0.050962300872197795,
      "grad_norm": 0.4512891229975294,
      "learning_rate": 0.0001997708774287068,
      "loss": 1.0841,
      "step": 707
    },
    {
      "epoch": 0.05103438333453471,
      "grad_norm": 1.0702416262685028,
      "learning_rate": 0.00019976929515754657,
      "loss": 1.2194,
      "step": 708
    },
    {
      "epoch": 0.05110646579687162,
      "grad_norm": 0.2659543390162145,
      "learning_rate": 0.00019976770744806725,
      "loss": 1.0746,
      "step": 709
    },
    {
      "epoch": 0.051178548259208534,
      "grad_norm": 0.3828741604513535,
      "learning_rate": 0.0001997661143003553,
      "loss": 1.1711,
      "step": 710
    },
    {
      "epoch": 0.05125063072154545,
      "grad_norm": 0.1688531825803737,
      "learning_rate": 0.0001997645157144976,
      "loss": 1.2171,
      "step": 711
    },
    {
      "epoch": 0.05132271318388236,
      "grad_norm": 0.17548733225124957,
      "learning_rate": 0.00019976291169058127,
      "loss": 1.1311,
      "step": 712
    },
    {
      "epoch": 0.05139479564621927,
      "grad_norm": 1.4506285762987536,
      "learning_rate": 0.00019976130222869375,
      "loss": 1.08,
      "step": 713
    },
    {
      "epoch": 0.051466878108556186,
      "grad_norm": 2.222680178275773,
      "learning_rate": 0.00019975968732892274,
      "loss": 1.1925,
      "step": 714
    },
    {
      "epoch": 0.0515389605708931,
      "grad_norm": 1.7708043412149488,
      "learning_rate": 0.0001997580669913563,
      "loss": 1.1629,
      "step": 715
    },
    {
      "epoch": 0.05161104303323002,
      "grad_norm": 0.22941787342297457,
      "learning_rate": 0.00019975644121608278,
      "loss": 1.1038,
      "step": 716
    },
    {
      "epoch": 0.05168312549556693,
      "grad_norm": 0.20540876085609938,
      "learning_rate": 0.00019975481000319073,
      "loss": 1.152,
      "step": 717
    },
    {
      "epoch": 0.051755207957903844,
      "grad_norm": 0.2541187569244646,
      "learning_rate": 0.00019975317335276915,
      "loss": 1.1556,
      "step": 718
    },
    {
      "epoch": 0.05182729042024076,
      "grad_norm": 0.1507589256701616,
      "learning_rate": 0.00019975153126490717,
      "loss": 1.1849,
      "step": 719
    },
    {
      "epoch": 0.05189937288257767,
      "grad_norm": 0.140439167044637,
      "learning_rate": 0.00019974988373969434,
      "loss": 1.2184,
      "step": 720
    },
    {
      "epoch": 0.05197145534491458,
      "grad_norm": 0.30839152582916785,
      "learning_rate": 0.00019974823077722044,
      "loss": 1.1735,
      "step": 721
    },
    {
      "epoch": 0.052043537807251496,
      "grad_norm": 0.4324292374857538,
      "learning_rate": 0.00019974657237757564,
      "loss": 1.0903,
      "step": 722
    },
    {
      "epoch": 0.05211562026958841,
      "grad_norm": 1.1788455975102823,
      "learning_rate": 0.00019974490854085026,
      "loss": 1.1387,
      "step": 723
    },
    {
      "epoch": 0.05218770273192532,
      "grad_norm": 0.9754175184428089,
      "learning_rate": 0.000199743239267135,
      "loss": 1.1381,
      "step": 724
    },
    {
      "epoch": 0.052259785194262234,
      "grad_norm": 0.30874298756648133,
      "learning_rate": 0.0001997415645565209,
      "loss": 1.1628,
      "step": 725
    },
    {
      "epoch": 0.05233186765659915,
      "grad_norm": 1.2256678332919768,
      "learning_rate": 0.0001997398844090992,
      "loss": 1.1149,
      "step": 726
    },
    {
      "epoch": 0.05240395011893606,
      "grad_norm": 0.21039293888243626,
      "learning_rate": 0.0001997381988249615,
      "loss": 1.0909,
      "step": 727
    },
    {
      "epoch": 0.05247603258127297,
      "grad_norm": 0.18619569336163413,
      "learning_rate": 0.00019973650780419968,
      "loss": 1.1453,
      "step": 728
    },
    {
      "epoch": 0.05254811504360989,
      "grad_norm": 0.18644379876505698,
      "learning_rate": 0.00019973481134690592,
      "loss": 1.1698,
      "step": 729
    },
    {
      "epoch": 0.052620197505946806,
      "grad_norm": 0.15134420771425988,
      "learning_rate": 0.00019973310945317271,
      "loss": 1.0908,
      "step": 730
    },
    {
      "epoch": 0.05269227996828372,
      "grad_norm": 0.205365561483798,
      "learning_rate": 0.00019973140212309277,
      "loss": 1.1281,
      "step": 731
    },
    {
      "epoch": 0.05276436243062063,
      "grad_norm": 0.5254932526643015,
      "learning_rate": 0.00019972968935675923,
      "loss": 1.1075,
      "step": 732
    },
    {
      "epoch": 0.052836444892957544,
      "grad_norm": 0.2805582410353221,
      "learning_rate": 0.00019972797115426537,
      "loss": 1.11,
      "step": 733
    },
    {
      "epoch": 0.05290852735529446,
      "grad_norm": 0.2155350689573949,
      "learning_rate": 0.0001997262475157049,
      "loss": 1.0912,
      "step": 734
    },
    {
      "epoch": 0.05298060981763137,
      "grad_norm": 0.6702290350556102,
      "learning_rate": 0.00019972451844117178,
      "loss": 1.1927,
      "step": 735
    },
    {
      "epoch": 0.05305269227996828,
      "grad_norm": 0.8264728790595776,
      "learning_rate": 0.00019972278393076023,
      "loss": 1.1347,
      "step": 736
    },
    {
      "epoch": 0.053124774742305196,
      "grad_norm": 0.9833148076698957,
      "learning_rate": 0.0001997210439845648,
      "loss": 1.1605,
      "step": 737
    },
    {
      "epoch": 0.05319685720464211,
      "grad_norm": 0.1832611253190944,
      "learning_rate": 0.00019971929860268035,
      "loss": 1.1263,
      "step": 738
    },
    {
      "epoch": 0.05326893966697902,
      "grad_norm": 0.5676876874232416,
      "learning_rate": 0.00019971754778520202,
      "loss": 1.2173,
      "step": 739
    },
    {
      "epoch": 0.053341022129315935,
      "grad_norm": 0.2142070578140735,
      "learning_rate": 0.00019971579153222523,
      "loss": 1.1204,
      "step": 740
    },
    {
      "epoch": 0.05341310459165285,
      "grad_norm": 0.13844110898048873,
      "learning_rate": 0.00019971402984384574,
      "loss": 1.1668,
      "step": 741
    },
    {
      "epoch": 0.05348518705398977,
      "grad_norm": 0.37309197784575454,
      "learning_rate": 0.00019971226272015954,
      "loss": 1.2207,
      "step": 742
    },
    {
      "epoch": 0.05355726951632668,
      "grad_norm": 1.600370422851834,
      "learning_rate": 0.00019971049016126298,
      "loss": 1.2293,
      "step": 743
    },
    {
      "epoch": 0.05362935197866359,
      "grad_norm": 1.5240335029244114,
      "learning_rate": 0.00019970871216725265,
      "loss": 1.0821,
      "step": 744
    },
    {
      "epoch": 0.053701434441000506,
      "grad_norm": 0.38919147917759006,
      "learning_rate": 0.0001997069287382255,
      "loss": 1.1916,
      "step": 745
    },
    {
      "epoch": 0.05377351690333742,
      "grad_norm": 1.2309400952388494,
      "learning_rate": 0.00019970513987427872,
      "loss": 1.1232,
      "step": 746
    },
    {
      "epoch": 0.05384559936567433,
      "grad_norm": 0.9770839751171941,
      "learning_rate": 0.00019970334557550983,
      "loss": 1.1442,
      "step": 747
    },
    {
      "epoch": 0.053917681828011245,
      "grad_norm": 0.14684079097733566,
      "learning_rate": 0.00019970154584201665,
      "loss": 1.1226,
      "step": 748
    },
    {
      "epoch": 0.05398976429034816,
      "grad_norm": 0.35160044538616025,
      "learning_rate": 0.00019969974067389728,
      "loss": 1.1504,
      "step": 749
    },
    {
      "epoch": 0.05406184675268507,
      "grad_norm": 0.1632141532713241,
      "learning_rate": 0.00019969793007125006,
      "loss": 1.1571,
      "step": 750
    },
    {
      "epoch": 0.054133929215021984,
      "grad_norm": 1.2254887525993963,
      "learning_rate": 0.00019969611403417375,
      "loss": 1.0681,
      "step": 751
    },
    {
      "epoch": 0.054206011677358897,
      "grad_norm": 0.1487895661973955,
      "learning_rate": 0.0001996942925627673,
      "loss": 1.0817,
      "step": 752
    },
    {
      "epoch": 0.05427809413969581,
      "grad_norm": 0.3018582203516395,
      "learning_rate": 0.00019969246565713005,
      "loss": 1.1642,
      "step": 753
    },
    {
      "epoch": 0.05435017660203272,
      "grad_norm": 0.4672482025910664,
      "learning_rate": 0.00019969063331736153,
      "loss": 1.1142,
      "step": 754
    },
    {
      "epoch": 0.05442225906436964,
      "grad_norm": 0.17440753757997698,
      "learning_rate": 0.00019968879554356164,
      "loss": 1.1075,
      "step": 755
    },
    {
      "epoch": 0.054494341526706555,
      "grad_norm": 1.1930039623333242,
      "learning_rate": 0.00019968695233583056,
      "loss": 1.1413,
      "step": 756
    },
    {
      "epoch": 0.05456642398904347,
      "grad_norm": 0.19848263277332243,
      "learning_rate": 0.0001996851036942687,
      "loss": 1.1098,
      "step": 757
    },
    {
      "epoch": 0.05463850645138038,
      "grad_norm": 0.2137718931955163,
      "learning_rate": 0.00019968324961897694,
      "loss": 1.1146,
      "step": 758
    },
    {
      "epoch": 0.054710588913717294,
      "grad_norm": 0.23694277892601306,
      "learning_rate": 0.00019968139011005627,
      "loss": 1.2097,
      "step": 759
    },
    {
      "epoch": 0.05478267137605421,
      "grad_norm": 0.7602676853252031,
      "learning_rate": 0.00019967952516760807,
      "loss": 1.2695,
      "step": 760
    },
    {
      "epoch": 0.05485475383839112,
      "grad_norm": 0.16288380868720184,
      "learning_rate": 0.00019967765479173398,
      "loss": 1.1412,
      "step": 761
    },
    {
      "epoch": 0.05492683630072803,
      "grad_norm": 0.173078150330527,
      "learning_rate": 0.00019967577898253596,
      "loss": 1.0795,
      "step": 762
    },
    {
      "epoch": 0.054998918763064945,
      "grad_norm": 0.3230241099819331,
      "learning_rate": 0.00019967389774011625,
      "loss": 1.1411,
      "step": 763
    },
    {
      "epoch": 0.05507100122540186,
      "grad_norm": 0.44172417005581416,
      "learning_rate": 0.00019967201106457742,
      "loss": 1.1366,
      "step": 764
    },
    {
      "epoch": 0.05514308368773877,
      "grad_norm": 0.14151980102457354,
      "learning_rate": 0.0001996701189560223,
      "loss": 1.0864,
      "step": 765
    },
    {
      "epoch": 0.055215166150075684,
      "grad_norm": 0.18198328774367933,
      "learning_rate": 0.000199668221414554,
      "loss": 1.2294,
      "step": 766
    },
    {
      "epoch": 0.0552872486124126,
      "grad_norm": 0.4476343652387423,
      "learning_rate": 0.000199666318440276,
      "loss": 1.0799,
      "step": 767
    },
    {
      "epoch": 0.05535933107474952,
      "grad_norm": 0.1548124924191351,
      "learning_rate": 0.00019966441003329197,
      "loss": 1.0895,
      "step": 768
    },
    {
      "epoch": 0.05543141353708643,
      "grad_norm": 0.2032714961503939,
      "learning_rate": 0.000199662496193706,
      "loss": 1.2387,
      "step": 769
    },
    {
      "epoch": 0.05550349599942334,
      "grad_norm": 0.3100936260928172,
      "learning_rate": 0.0001996605769216224,
      "loss": 1.1082,
      "step": 770
    },
    {
      "epoch": 0.055575578461760256,
      "grad_norm": 0.9033081969894429,
      "learning_rate": 0.00019965865221714573,
      "loss": 1.164,
      "step": 771
    },
    {
      "epoch": 0.05564766092409717,
      "grad_norm": 0.3460543784146283,
      "learning_rate": 0.00019965672208038096,
      "loss": 1.0576,
      "step": 772
    },
    {
      "epoch": 0.05571974338643408,
      "grad_norm": 0.672587094151454,
      "learning_rate": 0.00019965478651143328,
      "loss": 1.1222,
      "step": 773
    },
    {
      "epoch": 0.055791825848770994,
      "grad_norm": 0.9730041107892623,
      "learning_rate": 0.00019965284551040817,
      "loss": 1.1891,
      "step": 774
    },
    {
      "epoch": 0.05586390831110791,
      "grad_norm": 0.17808057756595636,
      "learning_rate": 0.0001996508990774115,
      "loss": 1.1409,
      "step": 775
    },
    {
      "epoch": 0.05593599077344482,
      "grad_norm": 0.8407076860767169,
      "learning_rate": 0.00019964894721254933,
      "loss": 1.1585,
      "step": 776
    },
    {
      "epoch": 0.05600807323578173,
      "grad_norm": 0.2442235036907916,
      "learning_rate": 0.00019964698991592803,
      "loss": 1.1355,
      "step": 777
    },
    {
      "epoch": 0.056080155698118646,
      "grad_norm": 0.26344766662342645,
      "learning_rate": 0.00019964502718765429,
      "loss": 1.1153,
      "step": 778
    },
    {
      "epoch": 0.05615223816045556,
      "grad_norm": 0.2695194844070752,
      "learning_rate": 0.0001996430590278352,
      "loss": 1.1707,
      "step": 779
    },
    {
      "epoch": 0.05622432062279247,
      "grad_norm": 0.8885727577739397,
      "learning_rate": 0.00019964108543657786,
      "loss": 1.2192,
      "step": 780
    },
    {
      "epoch": 0.05629640308512939,
      "grad_norm": 0.20402975948457658,
      "learning_rate": 0.00019963910641399002,
      "loss": 1.0429,
      "step": 781
    },
    {
      "epoch": 0.056368485547466304,
      "grad_norm": 0.22404542578963635,
      "learning_rate": 0.00019963712196017946,
      "loss": 1.0258,
      "step": 782
    },
    {
      "epoch": 0.05644056800980322,
      "grad_norm": 0.6004653821672248,
      "learning_rate": 0.00019963513207525436,
      "loss": 1.1397,
      "step": 783
    },
    {
      "epoch": 0.05651265047214013,
      "grad_norm": 0.4038697397705888,
      "learning_rate": 0.0001996331367593232,
      "loss": 1.2127,
      "step": 784
    },
    {
      "epoch": 0.05658473293447704,
      "grad_norm": 0.3080734345992231,
      "learning_rate": 0.00019963113601249478,
      "loss": 1.1306,
      "step": 785
    },
    {
      "epoch": 0.056656815396813956,
      "grad_norm": 0.9146296186370612,
      "learning_rate": 0.0001996291298348781,
      "loss": 1.1515,
      "step": 786
    },
    {
      "epoch": 0.05672889785915087,
      "grad_norm": 0.13803971588567707,
      "learning_rate": 0.00019962711822658254,
      "loss": 1.1232,
      "step": 787
    },
    {
      "epoch": 0.05680098032148778,
      "grad_norm": 0.6050426289862025,
      "learning_rate": 0.00019962510118771772,
      "loss": 1.2562,
      "step": 788
    },
    {
      "epoch": 0.056873062783824695,
      "grad_norm": 0.14059815977617104,
      "learning_rate": 0.00019962307871839366,
      "loss": 1.0341,
      "step": 789
    },
    {
      "epoch": 0.05694514524616161,
      "grad_norm": 0.12253001045587357,
      "learning_rate": 0.00019962105081872055,
      "loss": 1.1379,
      "step": 790
    },
    {
      "epoch": 0.05701722770849852,
      "grad_norm": 0.8607483780453833,
      "learning_rate": 0.00019961901748880888,
      "loss": 1.0853,
      "step": 791
    },
    {
      "epoch": 0.05708931017083543,
      "grad_norm": 0.45078853667307306,
      "learning_rate": 0.0001996169787287696,
      "loss": 1.116,
      "step": 792
    },
    {
      "epoch": 0.057161392633172346,
      "grad_norm": 0.3216519469294412,
      "learning_rate": 0.00019961493453871375,
      "loss": 1.0902,
      "step": 793
    },
    {
      "epoch": 0.05723347509550926,
      "grad_norm": 0.16459853821156764,
      "learning_rate": 0.00019961288491875278,
      "loss": 1.0796,
      "step": 794
    },
    {
      "epoch": 0.05730555755784618,
      "grad_norm": 0.2920006060443404,
      "learning_rate": 0.00019961082986899844,
      "loss": 1.1316,
      "step": 795
    },
    {
      "epoch": 0.05737764002018309,
      "grad_norm": 0.27495511117444715,
      "learning_rate": 0.0001996087693895627,
      "loss": 0.9887,
      "step": 796
    },
    {
      "epoch": 0.057449722482520005,
      "grad_norm": 0.201576701468438,
      "learning_rate": 0.00019960670348055794,
      "loss": 1.0956,
      "step": 797
    },
    {
      "epoch": 0.05752180494485692,
      "grad_norm": 0.31882852652228577,
      "learning_rate": 0.00019960463214209673,
      "loss": 1.1409,
      "step": 798
    },
    {
      "epoch": 0.05759388740719383,
      "grad_norm": 0.29144289623491026,
      "learning_rate": 0.00019960255537429194,
      "loss": 1.1362,
      "step": 799
    },
    {
      "epoch": 0.057665969869530743,
      "grad_norm": 0.46813187182300137,
      "learning_rate": 0.00019960047317725684,
      "loss": 1.1473,
      "step": 800
    },
    {
      "epoch": 0.057738052331867656,
      "grad_norm": 0.3467288222530635,
      "learning_rate": 0.00019959838555110493,
      "loss": 1.0523,
      "step": 801
    },
    {
      "epoch": 0.05781013479420457,
      "grad_norm": 1.07498722430977,
      "learning_rate": 0.00019959629249594992,
      "loss": 1.1971,
      "step": 802
    },
    {
      "epoch": 0.05788221725654148,
      "grad_norm": 0.7269165423018816,
      "learning_rate": 0.000199594194011906,
      "loss": 1.1998,
      "step": 803
    },
    {
      "epoch": 0.057954299718878395,
      "grad_norm": 0.19137580855140027,
      "learning_rate": 0.0001995920900990875,
      "loss": 1.0875,
      "step": 804
    },
    {
      "epoch": 0.05802638218121531,
      "grad_norm": 0.13286968751366604,
      "learning_rate": 0.0001995899807576091,
      "loss": 1.1441,
      "step": 805
    },
    {
      "epoch": 0.05809846464355222,
      "grad_norm": 0.18882732738665348,
      "learning_rate": 0.0001995878659875858,
      "loss": 1.0991,
      "step": 806
    },
    {
      "epoch": 0.058170547105889134,
      "grad_norm": 0.2115364115409222,
      "learning_rate": 0.00019958574578913285,
      "loss": 1.1534,
      "step": 807
    },
    {
      "epoch": 0.058242629568226054,
      "grad_norm": 0.1773433791445181,
      "learning_rate": 0.00019958362016236585,
      "loss": 1.0923,
      "step": 808
    },
    {
      "epoch": 0.05831471203056297,
      "grad_norm": 0.46184392285196013,
      "learning_rate": 0.00019958148910740063,
      "loss": 1.2068,
      "step": 809
    },
    {
      "epoch": 0.05838679449289988,
      "grad_norm": 0.27593454311011595,
      "learning_rate": 0.0001995793526243534,
      "loss": 1.1426,
      "step": 810
    },
    {
      "epoch": 0.05845887695523679,
      "grad_norm": 0.43572876746847145,
      "learning_rate": 0.0001995772107133406,
      "loss": 1.1107,
      "step": 811
    },
    {
      "epoch": 0.058530959417573705,
      "grad_norm": 0.3469122280177944,
      "learning_rate": 0.00019957506337447894,
      "loss": 1.1684,
      "step": 812
    },
    {
      "epoch": 0.05860304187991062,
      "grad_norm": 0.18585459888347822,
      "learning_rate": 0.00019957291060788553,
      "loss": 1.125,
      "step": 813
    },
    {
      "epoch": 0.05867512434224753,
      "grad_norm": 0.2088136067728675,
      "learning_rate": 0.00019957075241367766,
      "loss": 1.1884,
      "step": 814
    },
    {
      "epoch": 0.058747206804584444,
      "grad_norm": 0.15845285522801789,
      "learning_rate": 0.00019956858879197303,
      "loss": 1.0438,
      "step": 815
    },
    {
      "epoch": 0.05881928926692136,
      "grad_norm": 0.1180092270949206,
      "learning_rate": 0.0001995664197428895,
      "loss": 1.0877,
      "step": 816
    },
    {
      "epoch": 0.05889137172925827,
      "grad_norm": 0.11688745004455442,
      "learning_rate": 0.00019956424526654538,
      "loss": 1.1513,
      "step": 817
    },
    {
      "epoch": 0.05896345419159518,
      "grad_norm": 0.29422200942440957,
      "learning_rate": 0.00019956206536305915,
      "loss": 1.1017,
      "step": 818
    },
    {
      "epoch": 0.059035536653932096,
      "grad_norm": 0.20943604518583359,
      "learning_rate": 0.00019955988003254966,
      "loss": 1.142,
      "step": 819
    },
    {
      "epoch": 0.05910761911626901,
      "grad_norm": 0.2152512751150483,
      "learning_rate": 0.00019955768927513602,
      "loss": 1.1468,
      "step": 820
    },
    {
      "epoch": 0.05917970157860593,
      "grad_norm": 0.3338885222411793,
      "learning_rate": 0.00019955549309093765,
      "loss": 1.1471,
      "step": 821
    },
    {
      "epoch": 0.05925178404094284,
      "grad_norm": 0.3205061232221324,
      "learning_rate": 0.00019955329148007425,
      "loss": 1.1081,
      "step": 822
    },
    {
      "epoch": 0.059323866503279754,
      "grad_norm": 0.19635260277579564,
      "learning_rate": 0.00019955108444266585,
      "loss": 1.1364,
      "step": 823
    },
    {
      "epoch": 0.05939594896561667,
      "grad_norm": 0.7149699369765503,
      "learning_rate": 0.0001995488719788327,
      "loss": 1.1537,
      "step": 824
    },
    {
      "epoch": 0.05946803142795358,
      "grad_norm": 0.14675208449311294,
      "learning_rate": 0.00019954665408869547,
      "loss": 1.089,
      "step": 825
    },
    {
      "epoch": 0.05954011389029049,
      "grad_norm": 0.16816126505276294,
      "learning_rate": 0.000199544430772375,
      "loss": 1.058,
      "step": 826
    },
    {
      "epoch": 0.059612196352627406,
      "grad_norm": 0.23494189265089388,
      "learning_rate": 0.00019954220202999252,
      "loss": 1.1333,
      "step": 827
    },
    {
      "epoch": 0.05968427881496432,
      "grad_norm": 0.17135652010087768,
      "learning_rate": 0.0001995399678616695,
      "loss": 1.1272,
      "step": 828
    },
    {
      "epoch": 0.05975636127730123,
      "grad_norm": 0.25519157894837957,
      "learning_rate": 0.00019953772826752773,
      "loss": 1.1215,
      "step": 829
    },
    {
      "epoch": 0.059828443739638144,
      "grad_norm": 0.23122569226886794,
      "learning_rate": 0.00019953548324768927,
      "loss": 1.0737,
      "step": 830
    },
    {
      "epoch": 0.05990052620197506,
      "grad_norm": 0.15731780587506006,
      "learning_rate": 0.00019953323280227648,
      "loss": 1.1167,
      "step": 831
    },
    {
      "epoch": 0.05997260866431197,
      "grad_norm": 0.2680296693797616,
      "learning_rate": 0.00019953097693141208,
      "loss": 1.0995,
      "step": 832
    },
    {
      "epoch": 0.06004469112664888,
      "grad_norm": 0.1370221248982358,
      "learning_rate": 0.000199528715635219,
      "loss": 1.0817,
      "step": 833
    },
    {
      "epoch": 0.0601167735889858,
      "grad_norm": 0.17543686744936166,
      "learning_rate": 0.0001995264489138205,
      "loss": 1.0957,
      "step": 834
    },
    {
      "epoch": 0.060188856051322716,
      "grad_norm": 0.5222335239861542,
      "learning_rate": 0.00019952417676734015,
      "loss": 1.0986,
      "step": 835
    },
    {
      "epoch": 0.06026093851365963,
      "grad_norm": 0.32037151293112875,
      "learning_rate": 0.0001995218991959018,
      "loss": 1.1172,
      "step": 836
    },
    {
      "epoch": 0.06033302097599654,
      "grad_norm": 0.14321055643180705,
      "learning_rate": 0.00019951961619962962,
      "loss": 1.1116,
      "step": 837
    },
    {
      "epoch": 0.060405103438333455,
      "grad_norm": 0.2830833395775667,
      "learning_rate": 0.000199517327778648,
      "loss": 1.179,
      "step": 838
    },
    {
      "epoch": 0.06047718590067037,
      "grad_norm": 1.2222011469426484,
      "learning_rate": 0.0001995150339330817,
      "loss": 1.1799,
      "step": 839
    },
    {
      "epoch": 0.06054926836300728,
      "grad_norm": 0.8099649289357166,
      "learning_rate": 0.0001995127346630558,
      "loss": 0.9974,
      "step": 840
    },
    {
      "epoch": 0.06062135082534419,
      "grad_norm": 0.48862609616307257,
      "learning_rate": 0.00019951042996869557,
      "loss": 1.1844,
      "step": 841
    },
    {
      "epoch": 0.060693433287681106,
      "grad_norm": 0.1367543592402403,
      "learning_rate": 0.00019950811985012666,
      "loss": 1.1765,
      "step": 842
    },
    {
      "epoch": 0.06076551575001802,
      "grad_norm": 0.14618436941391166,
      "learning_rate": 0.000199505804307475,
      "loss": 1.157,
      "step": 843
    },
    {
      "epoch": 0.06083759821235493,
      "grad_norm": 0.12495888289523181,
      "learning_rate": 0.0001995034833408668,
      "loss": 1.0782,
      "step": 844
    },
    {
      "epoch": 0.060909680674691845,
      "grad_norm": 0.1138844899881968,
      "learning_rate": 0.00019950115695042858,
      "loss": 1.1704,
      "step": 845
    },
    {
      "epoch": 0.06098176313702876,
      "grad_norm": 0.30222916495974017,
      "learning_rate": 0.00019949882513628712,
      "loss": 1.0345,
      "step": 846
    },
    {
      "epoch": 0.06105384559936568,
      "grad_norm": 0.2513099833140484,
      "learning_rate": 0.0001994964878985696,
      "loss": 1.063,
      "step": 847
    },
    {
      "epoch": 0.06112592806170259,
      "grad_norm": 0.15467203479949304,
      "learning_rate": 0.0001994941452374033,
      "loss": 1.1698,
      "step": 848
    },
    {
      "epoch": 0.0611980105240395,
      "grad_norm": 0.21359870595687305,
      "learning_rate": 0.00019949179715291604,
      "loss": 1.181,
      "step": 849
    },
    {
      "epoch": 0.061270092986376416,
      "grad_norm": 0.34592306217573704,
      "learning_rate": 0.00019948944364523572,
      "loss": 1.1195,
      "step": 850
    },
    {
      "epoch": 0.06134217544871333,
      "grad_norm": 0.19544455552207343,
      "learning_rate": 0.0001994870847144907,
      "loss": 1.1371,
      "step": 851
    },
    {
      "epoch": 0.06141425791105024,
      "grad_norm": 0.15907074706593957,
      "learning_rate": 0.00019948472036080949,
      "loss": 1.0132,
      "step": 852
    },
    {
      "epoch": 0.061486340373387155,
      "grad_norm": 0.1997216988150738,
      "learning_rate": 0.000199482350584321,
      "loss": 1.1005,
      "step": 853
    },
    {
      "epoch": 0.06155842283572407,
      "grad_norm": 0.2997698379663361,
      "learning_rate": 0.00019947997538515445,
      "loss": 1.2007,
      "step": 854
    },
    {
      "epoch": 0.06163050529806098,
      "grad_norm": 0.14062452054194094,
      "learning_rate": 0.00019947759476343922,
      "loss": 1.1571,
      "step": 855
    },
    {
      "epoch": 0.061702587760397894,
      "grad_norm": 0.1590722800005528,
      "learning_rate": 0.00019947520871930517,
      "loss": 1.0579,
      "step": 856
    },
    {
      "epoch": 0.06177467022273481,
      "grad_norm": 0.20843535368830918,
      "learning_rate": 0.00019947281725288227,
      "loss": 1.0988,
      "step": 857
    },
    {
      "epoch": 0.06184675268507172,
      "grad_norm": 0.3539633135315316,
      "learning_rate": 0.00019947042036430097,
      "loss": 1.1246,
      "step": 858
    },
    {
      "epoch": 0.06191883514740863,
      "grad_norm": 0.1615874572823667,
      "learning_rate": 0.00019946801805369182,
      "loss": 1.1937,
      "step": 859
    },
    {
      "epoch": 0.06199091760974555,
      "grad_norm": 0.19947044593414692,
      "learning_rate": 0.00019946561032118586,
      "loss": 1.0074,
      "step": 860
    },
    {
      "epoch": 0.062063000072082465,
      "grad_norm": 1.539430241766345,
      "learning_rate": 0.00019946319716691426,
      "loss": 1.084,
      "step": 861
    },
    {
      "epoch": 0.06213508253441938,
      "grad_norm": 0.3474686753235254,
      "learning_rate": 0.00019946077859100863,
      "loss": 1.194,
      "step": 862
    },
    {
      "epoch": 0.06220716499675629,
      "grad_norm": 0.20450259897601206,
      "learning_rate": 0.00019945835459360073,
      "loss": 1.1233,
      "step": 863
    },
    {
      "epoch": 0.062279247459093204,
      "grad_norm": 0.19383601981707596,
      "learning_rate": 0.00019945592517482273,
      "loss": 1.0876,
      "step": 864
    },
    {
      "epoch": 0.06235132992143012,
      "grad_norm": 0.14373172272675366,
      "learning_rate": 0.00019945349033480706,
      "loss": 1.0164,
      "step": 865
    },
    {
      "epoch": 0.06242341238376703,
      "grad_norm": 0.12939029854764045,
      "learning_rate": 0.00019945105007368645,
      "loss": 1.1064,
      "step": 866
    },
    {
      "epoch": 0.06249549484610394,
      "grad_norm": 0.30037681990829973,
      "learning_rate": 0.00019944860439159387,
      "loss": 1.1062,
      "step": 867
    },
    {
      "epoch": 0.06256757730844086,
      "grad_norm": 0.25837237243610306,
      "learning_rate": 0.00019944615328866268,
      "loss": 1.0249,
      "step": 868
    },
    {
      "epoch": 0.06263965977077778,
      "grad_norm": 0.2653372615372882,
      "learning_rate": 0.00019944369676502642,
      "loss": 1.0612,
      "step": 869
    },
    {
      "epoch": 0.06271174223311468,
      "grad_norm": 0.250553610696317,
      "learning_rate": 0.00019944123482081908,
      "loss": 1.1435,
      "step": 870
    },
    {
      "epoch": 0.0627838246954516,
      "grad_norm": 0.3429324497837315,
      "learning_rate": 0.0001994387674561748,
      "loss": 1.1377,
      "step": 871
    },
    {
      "epoch": 0.06285590715778851,
      "grad_norm": 0.18270455611052125,
      "learning_rate": 0.00019943629467122808,
      "loss": 1.1487,
      "step": 872
    },
    {
      "epoch": 0.06292798962012543,
      "grad_norm": 0.6808260463201291,
      "learning_rate": 0.0001994338164661137,
      "loss": 1.1119,
      "step": 873
    },
    {
      "epoch": 0.06300007208246233,
      "grad_norm": 0.16222600268687645,
      "learning_rate": 0.00019943133284096683,
      "loss": 1.0603,
      "step": 874
    },
    {
      "epoch": 0.06307215454479925,
      "grad_norm": 0.14367821700325384,
      "learning_rate": 0.00019942884379592273,
      "loss": 1.1345,
      "step": 875
    },
    {
      "epoch": 0.06314423700713616,
      "grad_norm": 0.20752691220667452,
      "learning_rate": 0.00019942634933111715,
      "loss": 1.1201,
      "step": 876
    },
    {
      "epoch": 0.06321631946947308,
      "grad_norm": 0.132451203669555,
      "learning_rate": 0.000199423849446686,
      "loss": 1.0831,
      "step": 877
    },
    {
      "epoch": 0.06328840193180998,
      "grad_norm": 0.1552891481453513,
      "learning_rate": 0.00019942134414276566,
      "loss": 1.0694,
      "step": 878
    },
    {
      "epoch": 0.0633604843941469,
      "grad_norm": 1.9246685190764021,
      "learning_rate": 0.00019941883341949255,
      "loss": 1.1661,
      "step": 879
    },
    {
      "epoch": 0.06343256685648382,
      "grad_norm": 0.30066397473716977,
      "learning_rate": 0.0001994163172770036,
      "loss": 1.0731,
      "step": 880
    },
    {
      "epoch": 0.06350464931882073,
      "grad_norm": 0.16329308571499784,
      "learning_rate": 0.00019941379571543596,
      "loss": 1.1397,
      "step": 881
    },
    {
      "epoch": 0.06357673178115765,
      "grad_norm": 0.18276996208205942,
      "learning_rate": 0.00019941126873492708,
      "loss": 1.0837,
      "step": 882
    },
    {
      "epoch": 0.06364881424349456,
      "grad_norm": 1.8320943168888641,
      "learning_rate": 0.0001994087363356147,
      "loss": 1.1284,
      "step": 883
    },
    {
      "epoch": 0.06372089670583148,
      "grad_norm": 0.19150255551772058,
      "learning_rate": 0.00019940619851763685,
      "loss": 1.1438,
      "step": 884
    },
    {
      "epoch": 0.06379297916816838,
      "grad_norm": 0.17262129370610585,
      "learning_rate": 0.00019940365528113186,
      "loss": 1.0629,
      "step": 885
    },
    {
      "epoch": 0.0638650616305053,
      "grad_norm": 0.2791032998956983,
      "learning_rate": 0.00019940110662623837,
      "loss": 1.1703,
      "step": 886
    },
    {
      "epoch": 0.06393714409284221,
      "grad_norm": 0.13740992759918907,
      "learning_rate": 0.0001993985525530953,
      "loss": 1.1667,
      "step": 887
    },
    {
      "epoch": 0.06400922655517913,
      "grad_norm": 0.14347692523299596,
      "learning_rate": 0.00019939599306184188,
      "loss": 1.133,
      "step": 888
    },
    {
      "epoch": 0.06408130901751603,
      "grad_norm": 0.22446127999674578,
      "learning_rate": 0.0001993934281526176,
      "loss": 1.1372,
      "step": 889
    },
    {
      "epoch": 0.06415339147985295,
      "grad_norm": 0.17835506335282794,
      "learning_rate": 0.0001993908578255623,
      "loss": 1.0763,
      "step": 890
    },
    {
      "epoch": 0.06422547394218986,
      "grad_norm": 0.419064507768503,
      "learning_rate": 0.00019938828208081606,
      "loss": 1.129,
      "step": 891
    },
    {
      "epoch": 0.06429755640452678,
      "grad_norm": 0.16687217166798946,
      "learning_rate": 0.0001993857009185193,
      "loss": 1.1379,
      "step": 892
    },
    {
      "epoch": 0.0643696388668637,
      "grad_norm": 0.17373494122791355,
      "learning_rate": 0.00019938311433881272,
      "loss": 1.1338,
      "step": 893
    },
    {
      "epoch": 0.0644417213292006,
      "grad_norm": 0.15232854781257926,
      "learning_rate": 0.0001993805223418373,
      "loss": 1.136,
      "step": 894
    },
    {
      "epoch": 0.06451380379153752,
      "grad_norm": 0.1334522749000408,
      "learning_rate": 0.0001993779249277343,
      "loss": 1.2421,
      "step": 895
    },
    {
      "epoch": 0.06458588625387443,
      "grad_norm": 0.4307974932527687,
      "learning_rate": 0.00019937532209664534,
      "loss": 1.215,
      "step": 896
    },
    {
      "epoch": 0.06465796871621135,
      "grad_norm": 0.14337524119209072,
      "learning_rate": 0.00019937271384871233,
      "loss": 1.0938,
      "step": 897
    },
    {
      "epoch": 0.06473005117854826,
      "grad_norm": 0.17980674260254134,
      "learning_rate": 0.00019937010018407735,
      "loss": 1.0477,
      "step": 898
    },
    {
      "epoch": 0.06480213364088518,
      "grad_norm": 0.1499835558164339,
      "learning_rate": 0.00019936748110288295,
      "loss": 1.1412,
      "step": 899
    },
    {
      "epoch": 0.06487421610322208,
      "grad_norm": 0.17509529254800255,
      "learning_rate": 0.00019936485660527187,
      "loss": 1.1921,
      "step": 900
    },
    {
      "epoch": 0.064946298565559,
      "grad_norm": 0.17934466233006358,
      "learning_rate": 0.00019936222669138714,
      "loss": 1.06,
      "step": 901
    },
    {
      "epoch": 0.06501838102789591,
      "grad_norm": 0.18405270581939182,
      "learning_rate": 0.00019935959136137214,
      "loss": 1.0998,
      "step": 902
    },
    {
      "epoch": 0.06509046349023283,
      "grad_norm": 2.228791801382877,
      "learning_rate": 0.00019935695061537053,
      "loss": 1.1166,
      "step": 903
    },
    {
      "epoch": 0.06516254595256973,
      "grad_norm": 0.17659562083619904,
      "learning_rate": 0.00019935430445352622,
      "loss": 1.1096,
      "step": 904
    },
    {
      "epoch": 0.06523462841490665,
      "grad_norm": 0.17632324125963042,
      "learning_rate": 0.00019935165287598348,
      "loss": 1.1118,
      "step": 905
    },
    {
      "epoch": 0.06530671087724357,
      "grad_norm": 0.25850706523566214,
      "learning_rate": 0.00019934899588288683,
      "loss": 1.0485,
      "step": 906
    },
    {
      "epoch": 0.06537879333958048,
      "grad_norm": 0.6849588229420763,
      "learning_rate": 0.00019934633347438107,
      "loss": 1.1329,
      "step": 907
    },
    {
      "epoch": 0.0654508758019174,
      "grad_norm": 0.14198077935971903,
      "learning_rate": 0.0001993436656506114,
      "loss": 1.0439,
      "step": 908
    },
    {
      "epoch": 0.0655229582642543,
      "grad_norm": 0.2836528485606521,
      "learning_rate": 0.0001993409924117232,
      "loss": 1.0626,
      "step": 909
    },
    {
      "epoch": 0.06559504072659122,
      "grad_norm": 0.11512030995330891,
      "learning_rate": 0.00019933831375786216,
      "loss": 1.1209,
      "step": 910
    },
    {
      "epoch": 0.06566712318892813,
      "grad_norm": 0.1909079959322667,
      "learning_rate": 0.0001993356296891743,
      "loss": 1.1054,
      "step": 911
    },
    {
      "epoch": 0.06573920565126505,
      "grad_norm": 2.0303218940254295,
      "learning_rate": 0.000199332940205806,
      "loss": 1.137,
      "step": 912
    },
    {
      "epoch": 0.06581128811360196,
      "grad_norm": 0.13651263064295435,
      "learning_rate": 0.00019933024530790377,
      "loss": 1.0384,
      "step": 913
    },
    {
      "epoch": 0.06588337057593888,
      "grad_norm": 0.12682951158814676,
      "learning_rate": 0.0001993275449956145,
      "loss": 1.1045,
      "step": 914
    },
    {
      "epoch": 0.06595545303827578,
      "grad_norm": 0.2964066274309152,
      "learning_rate": 0.00019932483926908548,
      "loss": 1.1165,
      "step": 915
    },
    {
      "epoch": 0.0660275355006127,
      "grad_norm": 4.952434336285497,
      "learning_rate": 0.0001993221281284641,
      "loss": 1.1322,
      "step": 916
    },
    {
      "epoch": 0.06609961796294961,
      "grad_norm": 0.19825116352414474,
      "learning_rate": 0.00019931941157389817,
      "loss": 1.062,
      "step": 917
    },
    {
      "epoch": 0.06617170042528653,
      "grad_norm": 0.23979651658581244,
      "learning_rate": 0.0001993166896055358,
      "loss": 1.0297,
      "step": 918
    },
    {
      "epoch": 0.06624378288762345,
      "grad_norm": 0.25606469850755487,
      "learning_rate": 0.00019931396222352529,
      "loss": 1.1086,
      "step": 919
    },
    {
      "epoch": 0.06631586534996035,
      "grad_norm": 0.3707368864036322,
      "learning_rate": 0.00019931122942801538,
      "loss": 1.1315,
      "step": 920
    },
    {
      "epoch": 0.06638794781229727,
      "grad_norm": 0.5361922962178773,
      "learning_rate": 0.000199308491219155,
      "loss": 1.1121,
      "step": 921
    },
    {
      "epoch": 0.06646003027463418,
      "grad_norm": 1.2944043881317997,
      "learning_rate": 0.0001993057475970934,
      "loss": 1.0836,
      "step": 922
    },
    {
      "epoch": 0.0665321127369711,
      "grad_norm": 0.9158688690542139,
      "learning_rate": 0.00019930299856198017,
      "loss": 1.1285,
      "step": 923
    },
    {
      "epoch": 0.066604195199308,
      "grad_norm": 0.8430930166052738,
      "learning_rate": 0.00019930024411396512,
      "loss": 1.0871,
      "step": 924
    },
    {
      "epoch": 0.06667627766164493,
      "grad_norm": 2.069723191947447,
      "learning_rate": 0.00019929748425319839,
      "loss": 1.0809,
      "step": 925
    },
    {
      "epoch": 0.06674836012398183,
      "grad_norm": 0.26662950530827506,
      "learning_rate": 0.00019929471897983043,
      "loss": 1.0932,
      "step": 926
    },
    {
      "epoch": 0.06682044258631875,
      "grad_norm": 0.5314809014819708,
      "learning_rate": 0.00019929194829401196,
      "loss": 1.2728,
      "step": 927
    },
    {
      "epoch": 0.06689252504865566,
      "grad_norm": 0.2733993350697218,
      "learning_rate": 0.00019928917219589403,
      "loss": 1.1335,
      "step": 928
    },
    {
      "epoch": 0.06696460751099258,
      "grad_norm": 0.1955110458689088,
      "learning_rate": 0.00019928639068562794,
      "loss": 1.1188,
      "step": 929
    },
    {
      "epoch": 0.06703668997332948,
      "grad_norm": 0.15213572078958262,
      "learning_rate": 0.00019928360376336534,
      "loss": 1.1327,
      "step": 930
    },
    {
      "epoch": 0.0671087724356664,
      "grad_norm": 0.17615074043071915,
      "learning_rate": 0.00019928081142925807,
      "loss": 1.1247,
      "step": 931
    },
    {
      "epoch": 0.06718085489800332,
      "grad_norm": 0.2575977350647759,
      "learning_rate": 0.0001992780136834584,
      "loss": 1.1351,
      "step": 932
    },
    {
      "epoch": 0.06725293736034023,
      "grad_norm": 0.5701501192319849,
      "learning_rate": 0.00019927521052611884,
      "loss": 1.0757,
      "step": 933
    },
    {
      "epoch": 0.06732501982267715,
      "grad_norm": 0.2282498647334421,
      "learning_rate": 0.00019927240195739216,
      "loss": 1.0954,
      "step": 934
    },
    {
      "epoch": 0.06739710228501405,
      "grad_norm": 0.13993630602652773,
      "learning_rate": 0.00019926958797743144,
      "loss": 0.9976,
      "step": 935
    },
    {
      "epoch": 0.06746918474735097,
      "grad_norm": 0.2640512291432146,
      "learning_rate": 0.00019926676858639011,
      "loss": 1.1553,
      "step": 936
    },
    {
      "epoch": 0.06754126720968788,
      "grad_norm": 3.522710170393808,
      "learning_rate": 0.00019926394378442182,
      "loss": 1.1811,
      "step": 937
    },
    {
      "epoch": 0.0676133496720248,
      "grad_norm": 0.20358788011174178,
      "learning_rate": 0.00019926111357168055,
      "loss": 1.1463,
      "step": 938
    },
    {
      "epoch": 0.0676854321343617,
      "grad_norm": 0.6428669117075281,
      "learning_rate": 0.00019925827794832056,
      "loss": 1.1559,
      "step": 939
    },
    {
      "epoch": 0.06775751459669863,
      "grad_norm": 0.14301022784097617,
      "learning_rate": 0.00019925543691449643,
      "loss": 1.1084,
      "step": 940
    },
    {
      "epoch": 0.06782959705903553,
      "grad_norm": 0.3090789975083523,
      "learning_rate": 0.00019925259047036303,
      "loss": 1.0298,
      "step": 941
    },
    {
      "epoch": 0.06790167952137245,
      "grad_norm": 2.7223296587674515,
      "learning_rate": 0.00019924973861607553,
      "loss": 1.0667,
      "step": 942
    },
    {
      "epoch": 0.06797376198370936,
      "grad_norm": 0.593457927177477,
      "learning_rate": 0.00019924688135178936,
      "loss": 1.0921,
      "step": 943
    },
    {
      "epoch": 0.06804584444604628,
      "grad_norm": 0.540191386293367,
      "learning_rate": 0.00019924401867766025,
      "loss": 1.118,
      "step": 944
    },
    {
      "epoch": 0.0681179269083832,
      "grad_norm": 0.16843450766739054,
      "learning_rate": 0.00019924115059384424,
      "loss": 1.1527,
      "step": 945
    },
    {
      "epoch": 0.0681900093707201,
      "grad_norm": 1.8057992067228104,
      "learning_rate": 0.00019923827710049773,
      "loss": 1.1421,
      "step": 946
    },
    {
      "epoch": 0.06826209183305702,
      "grad_norm": 1.5713800011051031,
      "learning_rate": 0.0001992353981977773,
      "loss": 1.194,
      "step": 947
    },
    {
      "epoch": 0.06833417429539393,
      "grad_norm": 0.2772261193842621,
      "learning_rate": 0.00019923251388583987,
      "loss": 1.2468,
      "step": 948
    },
    {
      "epoch": 0.06840625675773085,
      "grad_norm": 0.20658475810957047,
      "learning_rate": 0.00019922962416484268,
      "loss": 1.1443,
      "step": 949
    },
    {
      "epoch": 0.06847833922006775,
      "grad_norm": 0.29760877601242963,
      "learning_rate": 0.00019922672903494325,
      "loss": 1.1594,
      "step": 950
    },
    {
      "epoch": 0.06855042168240467,
      "grad_norm": 0.5522537884298538,
      "learning_rate": 0.00019922382849629934,
      "loss": 1.1153,
      "step": 951
    },
    {
      "epoch": 0.06862250414474158,
      "grad_norm": 1.0766155660197998,
      "learning_rate": 0.00019922092254906915,
      "loss": 1.2366,
      "step": 952
    },
    {
      "epoch": 0.0686945866070785,
      "grad_norm": 1.1561896109484118,
      "learning_rate": 0.000199218011193411,
      "loss": 1.2009,
      "step": 953
    },
    {
      "epoch": 0.0687666690694154,
      "grad_norm": 0.22862795993229934,
      "learning_rate": 0.00019921509442948358,
      "loss": 1.1402,
      "step": 954
    },
    {
      "epoch": 0.06883875153175233,
      "grad_norm": 0.46001394822094505,
      "learning_rate": 0.00019921217225744595,
      "loss": 1.0809,
      "step": 955
    },
    {
      "epoch": 0.06891083399408923,
      "grad_norm": 0.7723327506067476,
      "learning_rate": 0.00019920924467745733,
      "loss": 1.1005,
      "step": 956
    },
    {
      "epoch": 0.06898291645642615,
      "grad_norm": 1.6431836160242697,
      "learning_rate": 0.00019920631168967733,
      "loss": 1.1198,
      "step": 957
    },
    {
      "epoch": 0.06905499891876306,
      "grad_norm": 0.18822346943300639,
      "learning_rate": 0.00019920337329426584,
      "loss": 1.055,
      "step": 958
    },
    {
      "epoch": 0.06912708138109998,
      "grad_norm": 0.22899878883164984,
      "learning_rate": 0.000199200429491383,
      "loss": 1.1178,
      "step": 959
    },
    {
      "epoch": 0.0691991638434369,
      "grad_norm": 0.24341539906133522,
      "learning_rate": 0.00019919748028118925,
      "loss": 1.2103,
      "step": 960
    },
    {
      "epoch": 0.0692712463057738,
      "grad_norm": 0.4225241979955316,
      "learning_rate": 0.0001991945256638454,
      "loss": 1.0922,
      "step": 961
    },
    {
      "epoch": 0.06934332876811072,
      "grad_norm": 0.1390805238230586,
      "learning_rate": 0.00019919156563951248,
      "loss": 1.0776,
      "step": 962
    },
    {
      "epoch": 0.06941541123044763,
      "grad_norm": 0.15802661749679198,
      "learning_rate": 0.00019918860020835184,
      "loss": 1.1202,
      "step": 963
    },
    {
      "epoch": 0.06948749369278455,
      "grad_norm": 0.18312848429894146,
      "learning_rate": 0.00019918562937052512,
      "loss": 1.1191,
      "step": 964
    },
    {
      "epoch": 0.06955957615512146,
      "grad_norm": 0.1349953450399105,
      "learning_rate": 0.00019918265312619427,
      "loss": 1.0796,
      "step": 965
    },
    {
      "epoch": 0.06963165861745838,
      "grad_norm": 0.6051683401738657,
      "learning_rate": 0.0001991796714755215,
      "loss": 1.0623,
      "step": 966
    },
    {
      "epoch": 0.06970374107979528,
      "grad_norm": 2.611045676013433,
      "learning_rate": 0.00019917668441866935,
      "loss": 1.0836,
      "step": 967
    },
    {
      "epoch": 0.0697758235421322,
      "grad_norm": 0.14212565618283626,
      "learning_rate": 0.00019917369195580063,
      "loss": 1.1115,
      "step": 968
    },
    {
      "epoch": 0.0698479060044691,
      "grad_norm": 0.8633481849763797,
      "learning_rate": 0.00019917069408707848,
      "loss": 1.1878,
      "step": 969
    },
    {
      "epoch": 0.06991998846680603,
      "grad_norm": 0.4802142281606951,
      "learning_rate": 0.00019916769081266628,
      "loss": 1.1084,
      "step": 970
    },
    {
      "epoch": 0.06999207092914293,
      "grad_norm": 0.8263783399666299,
      "learning_rate": 0.00019916468213272776,
      "loss": 1.017,
      "step": 971
    },
    {
      "epoch": 0.07006415339147985,
      "grad_norm": 0.681311986019335,
      "learning_rate": 0.00019916166804742692,
      "loss": 1.2464,
      "step": 972
    },
    {
      "epoch": 0.07013623585381677,
      "grad_norm": 0.46832022664704037,
      "learning_rate": 0.00019915864855692803,
      "loss": 1.1257,
      "step": 973
    },
    {
      "epoch": 0.07020831831615368,
      "grad_norm": 0.3693017678639236,
      "learning_rate": 0.0001991556236613957,
      "loss": 1.189,
      "step": 974
    },
    {
      "epoch": 0.0702804007784906,
      "grad_norm": 0.796989271494132,
      "learning_rate": 0.00019915259336099483,
      "loss": 1.0739,
      "step": 975
    },
    {
      "epoch": 0.0703524832408275,
      "grad_norm": 0.9671866828635577,
      "learning_rate": 0.00019914955765589057,
      "loss": 1.1474,
      "step": 976
    },
    {
      "epoch": 0.07042456570316442,
      "grad_norm": 1.0062356219331503,
      "learning_rate": 0.00019914651654624838,
      "loss": 1.0648,
      "step": 977
    },
    {
      "epoch": 0.07049664816550133,
      "grad_norm": 0.2732652684319651,
      "learning_rate": 0.00019914347003223407,
      "loss": 1.1066,
      "step": 978
    },
    {
      "epoch": 0.07056873062783825,
      "grad_norm": 1.469341647431488,
      "learning_rate": 0.00019914041811401365,
      "loss": 1.0959,
      "step": 979
    },
    {
      "epoch": 0.07064081309017516,
      "grad_norm": 0.3270651676700753,
      "learning_rate": 0.00019913736079175354,
      "loss": 1.0551,
      "step": 980
    },
    {
      "epoch": 0.07071289555251208,
      "grad_norm": 0.3801343728074387,
      "learning_rate": 0.00019913429806562034,
      "loss": 1.0968,
      "step": 981
    },
    {
      "epoch": 0.07078497801484898,
      "grad_norm": 0.39614275346866973,
      "learning_rate": 0.00019913122993578104,
      "loss": 1.0775,
      "step": 982
    },
    {
      "epoch": 0.0708570604771859,
      "grad_norm": 0.24330026865447438,
      "learning_rate": 0.00019912815640240282,
      "loss": 1.1564,
      "step": 983
    },
    {
      "epoch": 0.07092914293952281,
      "grad_norm": 0.23658744788676317,
      "learning_rate": 0.00019912507746565328,
      "loss": 1.1423,
      "step": 984
    },
    {
      "epoch": 0.07100122540185973,
      "grad_norm": 0.3067425081867489,
      "learning_rate": 0.0001991219931257002,
      "loss": 1.0094,
      "step": 985
    },
    {
      "epoch": 0.07107330786419665,
      "grad_norm": 0.28524100314970247,
      "learning_rate": 0.0001991189033827117,
      "loss": 1.1249,
      "step": 986
    },
    {
      "epoch": 0.07114539032653355,
      "grad_norm": 0.9240450216440788,
      "learning_rate": 0.00019911580823685627,
      "loss": 1.1461,
      "step": 987
    },
    {
      "epoch": 0.07121747278887047,
      "grad_norm": 0.17558496636036186,
      "learning_rate": 0.00019911270768830256,
      "loss": 1.1242,
      "step": 988
    },
    {
      "epoch": 0.07128955525120738,
      "grad_norm": 0.12102246081630146,
      "learning_rate": 0.0001991096017372196,
      "loss": 1.1031,
      "step": 989
    },
    {
      "epoch": 0.0713616377135443,
      "grad_norm": 0.42511122552508956,
      "learning_rate": 0.00019910649038377666,
      "loss": 1.0931,
      "step": 990
    },
    {
      "epoch": 0.0714337201758812,
      "grad_norm": 0.3463720454344734,
      "learning_rate": 0.0001991033736281434,
      "loss": 1.1962,
      "step": 991
    },
    {
      "epoch": 0.07150580263821812,
      "grad_norm": 1.7623614226305715,
      "learning_rate": 0.00019910025147048963,
      "loss": 1.186,
      "step": 992
    },
    {
      "epoch": 0.07157788510055503,
      "grad_norm": 2.1095722396120076,
      "learning_rate": 0.0001990971239109856,
      "loss": 1.0509,
      "step": 993
    },
    {
      "epoch": 0.07164996756289195,
      "grad_norm": 0.26652380042537505,
      "learning_rate": 0.00019909399094980178,
      "loss": 1.1675,
      "step": 994
    },
    {
      "epoch": 0.07172205002522886,
      "grad_norm": 0.3430919164991887,
      "learning_rate": 0.00019909085258710892,
      "loss": 1.0956,
      "step": 995
    },
    {
      "epoch": 0.07179413248756578,
      "grad_norm": 0.45434752506068915,
      "learning_rate": 0.0001990877088230781,
      "loss": 1.0716,
      "step": 996
    },
    {
      "epoch": 0.07186621494990268,
      "grad_norm": 0.532892310694834,
      "learning_rate": 0.00019908455965788067,
      "loss": 1.103,
      "step": 997
    },
    {
      "epoch": 0.0719382974122396,
      "grad_norm": 0.34985147388942583,
      "learning_rate": 0.00019908140509168834,
      "loss": 1.1385,
      "step": 998
    },
    {
      "epoch": 0.07201037987457652,
      "grad_norm": 0.5024557739660986,
      "learning_rate": 0.00019907824512467302,
      "loss": 1.1321,
      "step": 999
    },
    {
      "epoch": 0.07208246233691343,
      "grad_norm": 0.24698278780674507,
      "learning_rate": 0.00019907507975700693,
      "loss": 1.1574,
      "step": 1000
    },
    {
      "epoch": 0.07215454479925035,
      "grad_norm": 2.044796491384113,
      "learning_rate": 0.0001990719089888627,
      "loss": 1.0866,
      "step": 1001
    },
    {
      "epoch": 0.07222662726158725,
      "grad_norm": 0.23982420501245438,
      "learning_rate": 0.00019906873282041302,
      "loss": 1.0582,
      "step": 1002
    },
    {
      "epoch": 0.07229870972392417,
      "grad_norm": 0.9138451018040735,
      "learning_rate": 0.00019906555125183116,
      "loss": 1.1858,
      "step": 1003
    },
    {
      "epoch": 0.07237079218626108,
      "grad_norm": 1.883561712746871,
      "learning_rate": 0.0001990623642832905,
      "loss": 1.1603,
      "step": 1004
    },
    {
      "epoch": 0.072442874648598,
      "grad_norm": 0.5607764732021211,
      "learning_rate": 0.00019905917191496473,
      "loss": 1.0298,
      "step": 1005
    },
    {
      "epoch": 0.0725149571109349,
      "grad_norm": 0.9006134335608763,
      "learning_rate": 0.0001990559741470279,
      "loss": 1.1245,
      "step": 1006
    },
    {
      "epoch": 0.07258703957327182,
      "grad_norm": 1.2874019323126793,
      "learning_rate": 0.00019905277097965427,
      "loss": 1.1876,
      "step": 1007
    },
    {
      "epoch": 0.07265912203560873,
      "grad_norm": 0.21790652585222894,
      "learning_rate": 0.0001990495624130185,
      "loss": 1.1324,
      "step": 1008
    },
    {
      "epoch": 0.07273120449794565,
      "grad_norm": 1.548705370766249,
      "learning_rate": 0.00019904634844729544,
      "loss": 1.1657,
      "step": 1009
    },
    {
      "epoch": 0.07280328696028256,
      "grad_norm": 0.26221320983000607,
      "learning_rate": 0.0001990431290826603,
      "loss": 1.1431,
      "step": 1010
    },
    {
      "epoch": 0.07287536942261948,
      "grad_norm": 0.14861563359396687,
      "learning_rate": 0.00019903990431928854,
      "loss": 1.1105,
      "step": 1011
    },
    {
      "epoch": 0.0729474518849564,
      "grad_norm": 0.2498988546951299,
      "learning_rate": 0.00019903667415735598,
      "loss": 1.1094,
      "step": 1012
    },
    {
      "epoch": 0.0730195343472933,
      "grad_norm": 0.602717418695394,
      "learning_rate": 0.00019903343859703867,
      "loss": 1.1401,
      "step": 1013
    },
    {
      "epoch": 0.07309161680963022,
      "grad_norm": 0.5098922497417958,
      "learning_rate": 0.00019903019763851297,
      "loss": 1.1069,
      "step": 1014
    },
    {
      "epoch": 0.07316369927196713,
      "grad_norm": 0.2503556828114115,
      "learning_rate": 0.00019902695128195554,
      "loss": 1.1014,
      "step": 1015
    },
    {
      "epoch": 0.07323578173430405,
      "grad_norm": 0.5272744943631861,
      "learning_rate": 0.00019902369952754336,
      "loss": 1.0507,
      "step": 1016
    },
    {
      "epoch": 0.07330786419664095,
      "grad_norm": 0.26624001435496353,
      "learning_rate": 0.00019902044237545368,
      "loss": 1.0966,
      "step": 1017
    },
    {
      "epoch": 0.07337994665897787,
      "grad_norm": 0.30525213988357924,
      "learning_rate": 0.000199017179825864,
      "loss": 1.1354,
      "step": 1018
    },
    {
      "epoch": 0.07345202912131478,
      "grad_norm": 0.6513479931574266,
      "learning_rate": 0.0001990139118789522,
      "loss": 1.1354,
      "step": 1019
    },
    {
      "epoch": 0.0735241115836517,
      "grad_norm": 0.14544651809655235,
      "learning_rate": 0.0001990106385348964,
      "loss": 1.0778,
      "step": 1020
    },
    {
      "epoch": 0.0735961940459886,
      "grad_norm": 0.43439502537036706,
      "learning_rate": 0.00019900735979387505,
      "loss": 1.192,
      "step": 1021
    },
    {
      "epoch": 0.07366827650832553,
      "grad_norm": 1.039030610042117,
      "learning_rate": 0.0001990040756560668,
      "loss": 1.1038,
      "step": 1022
    },
    {
      "epoch": 0.07374035897066243,
      "grad_norm": 2.1148902241086485,
      "learning_rate": 0.00019900078612165075,
      "loss": 1.0193,
      "step": 1023
    },
    {
      "epoch": 0.07381244143299935,
      "grad_norm": 0.41377486092607907,
      "learning_rate": 0.00019899749119080612,
      "loss": 1.172,
      "step": 1024
    },
    {
      "epoch": 0.07388452389533627,
      "grad_norm": 1.1500908053588244,
      "learning_rate": 0.0001989941908637126,
      "loss": 1.109,
      "step": 1025
    },
    {
      "epoch": 0.07395660635767318,
      "grad_norm": 0.19844495644969995,
      "learning_rate": 0.00019899088514055004,
      "loss": 1.127,
      "step": 1026
    },
    {
      "epoch": 0.0740286888200101,
      "grad_norm": 0.641338596196738,
      "learning_rate": 0.00019898757402149863,
      "loss": 1.0837,
      "step": 1027
    },
    {
      "epoch": 0.074100771282347,
      "grad_norm": 2.152889573995206,
      "learning_rate": 0.0001989842575067389,
      "loss": 1.1504,
      "step": 1028
    },
    {
      "epoch": 0.07417285374468392,
      "grad_norm": 0.13384066021449526,
      "learning_rate": 0.0001989809355964516,
      "loss": 1.1474,
      "step": 1029
    },
    {
      "epoch": 0.07424493620702083,
      "grad_norm": 1.2773006853459117,
      "learning_rate": 0.00019897760829081778,
      "loss": 1.1103,
      "step": 1030
    },
    {
      "epoch": 0.07431701866935775,
      "grad_norm": 0.6044467762860439,
      "learning_rate": 0.0001989742755900188,
      "loss": 1.1067,
      "step": 1031
    },
    {
      "epoch": 0.07438910113169465,
      "grad_norm": 0.2678689248001368,
      "learning_rate": 0.00019897093749423638,
      "loss": 1.1397,
      "step": 1032
    },
    {
      "epoch": 0.07446118359403157,
      "grad_norm": 0.28882920960496766,
      "learning_rate": 0.00019896759400365248,
      "loss": 1.1369,
      "step": 1033
    },
    {
      "epoch": 0.07453326605636848,
      "grad_norm": 0.2535136854840574,
      "learning_rate": 0.00019896424511844926,
      "loss": 1.1509,
      "step": 1034
    },
    {
      "epoch": 0.0746053485187054,
      "grad_norm": 0.25675227437808984,
      "learning_rate": 0.00019896089083880936,
      "loss": 1.1035,
      "step": 1035
    },
    {
      "epoch": 0.0746774309810423,
      "grad_norm": 0.4351383177223063,
      "learning_rate": 0.00019895753116491559,
      "loss": 1.1767,
      "step": 1036
    },
    {
      "epoch": 0.07474951344337923,
      "grad_norm": 0.1920550101761818,
      "learning_rate": 0.00019895416609695105,
      "loss": 1.0497,
      "step": 1037
    },
    {
      "epoch": 0.07482159590571615,
      "grad_norm": 1.708474147594327,
      "learning_rate": 0.00019895079563509916,
      "loss": 1.1236,
      "step": 1038
    },
    {
      "epoch": 0.07489367836805305,
      "grad_norm": 0.4089117672005981,
      "learning_rate": 0.0001989474197795437,
      "loss": 1.1562,
      "step": 1039
    },
    {
      "epoch": 0.07496576083038997,
      "grad_norm": 0.11190187584991698,
      "learning_rate": 0.00019894403853046866,
      "loss": 1.0901,
      "step": 1040
    },
    {
      "epoch": 0.07503784329272688,
      "grad_norm": 0.3672102604560194,
      "learning_rate": 0.00019894065188805833,
      "loss": 1.0652,
      "step": 1041
    },
    {
      "epoch": 0.0751099257550638,
      "grad_norm": 0.47893613807664676,
      "learning_rate": 0.00019893725985249732,
      "loss": 1.1059,
      "step": 1042
    },
    {
      "epoch": 0.0751820082174007,
      "grad_norm": 0.6177315713318786,
      "learning_rate": 0.00019893386242397055,
      "loss": 1.1424,
      "step": 1043
    },
    {
      "epoch": 0.07525409067973762,
      "grad_norm": 0.24998542196488752,
      "learning_rate": 0.00019893045960266316,
      "loss": 1.1048,
      "step": 1044
    },
    {
      "epoch": 0.07532617314207453,
      "grad_norm": 0.2327000272800188,
      "learning_rate": 0.00019892705138876068,
      "loss": 1.2058,
      "step": 1045
    },
    {
      "epoch": 0.07539825560441145,
      "grad_norm": 0.8756086492008259,
      "learning_rate": 0.00019892363778244886,
      "loss": 1.1741,
      "step": 1046
    },
    {
      "epoch": 0.07547033806674835,
      "grad_norm": 0.25052938694991134,
      "learning_rate": 0.00019892021878391384,
      "loss": 1.1316,
      "step": 1047
    },
    {
      "epoch": 0.07554242052908527,
      "grad_norm": 1.7174940881610814,
      "learning_rate": 0.00019891679439334187,
      "loss": 1.2125,
      "step": 1048
    },
    {
      "epoch": 0.07561450299142218,
      "grad_norm": 0.4650081069468822,
      "learning_rate": 0.00019891336461091966,
      "loss": 1.172,
      "step": 1049
    },
    {
      "epoch": 0.0756865854537591,
      "grad_norm": 0.32586445703985156,
      "learning_rate": 0.0001989099294368342,
      "loss": 1.1228,
      "step": 1050
    },
    {
      "epoch": 0.07575866791609602,
      "grad_norm": 1.854599721284622,
      "learning_rate": 0.00019890648887127275,
      "loss": 1.1105,
      "step": 1051
    },
    {
      "epoch": 0.07583075037843293,
      "grad_norm": 1.5516570712137983,
      "learning_rate": 0.00019890304291442278,
      "loss": 1.0673,
      "step": 1052
    },
    {
      "epoch": 0.07590283284076985,
      "grad_norm": 0.15098141086585412,
      "learning_rate": 0.00019889959156647215,
      "loss": 1.073,
      "step": 1053
    },
    {
      "epoch": 0.07597491530310675,
      "grad_norm": 0.1461120163896009,
      "learning_rate": 0.000198896134827609,
      "loss": 1.1634,
      "step": 1054
    },
    {
      "epoch": 0.07604699776544367,
      "grad_norm": 0.39405748648115496,
      "learning_rate": 0.00019889267269802176,
      "loss": 1.0581,
      "step": 1055
    },
    {
      "epoch": 0.07611908022778058,
      "grad_norm": 0.4398243043313667,
      "learning_rate": 0.00019888920517789914,
      "loss": 1.1174,
      "step": 1056
    },
    {
      "epoch": 0.0761911626901175,
      "grad_norm": 0.3116766566596904,
      "learning_rate": 0.00019888573226743014,
      "loss": 1.1799,
      "step": 1057
    },
    {
      "epoch": 0.0762632451524544,
      "grad_norm": 0.1257585582303349,
      "learning_rate": 0.0001988822539668041,
      "loss": 1.0648,
      "step": 1058
    },
    {
      "epoch": 0.07633532761479132,
      "grad_norm": 0.22171754073986966,
      "learning_rate": 0.0001988787702762106,
      "loss": 1.1629,
      "step": 1059
    },
    {
      "epoch": 0.07640741007712823,
      "grad_norm": 0.254425906865195,
      "learning_rate": 0.00019887528119583948,
      "loss": 1.1788,
      "step": 1060
    },
    {
      "epoch": 0.07647949253946515,
      "grad_norm": 0.15592040491051903,
      "learning_rate": 0.000198871786725881,
      "loss": 1.2027,
      "step": 1061
    },
    {
      "epoch": 0.07655157500180206,
      "grad_norm": 0.1102017693274922,
      "learning_rate": 0.0001988682868665256,
      "loss": 1.2012,
      "step": 1062
    },
    {
      "epoch": 0.07662365746413898,
      "grad_norm": 0.19684334940190976,
      "learning_rate": 0.00019886478161796408,
      "loss": 1.2135,
      "step": 1063
    },
    {
      "epoch": 0.0766957399264759,
      "grad_norm": 0.3462453987425941,
      "learning_rate": 0.00019886127098038745,
      "loss": 1.1331,
      "step": 1064
    },
    {
      "epoch": 0.0767678223888128,
      "grad_norm": 0.23915434172893946,
      "learning_rate": 0.00019885775495398714,
      "loss": 1.2049,
      "step": 1065
    },
    {
      "epoch": 0.07683990485114972,
      "grad_norm": 0.17250658219814902,
      "learning_rate": 0.0001988542335389548,
      "loss": 1.0775,
      "step": 1066
    },
    {
      "epoch": 0.07691198731348663,
      "grad_norm": 0.22150135886326305,
      "learning_rate": 0.00019885070673548235,
      "loss": 1.1463,
      "step": 1067
    },
    {
      "epoch": 0.07698406977582355,
      "grad_norm": 0.1321971194500292,
      "learning_rate": 0.000198847174543762,
      "loss": 1.1337,
      "step": 1068
    },
    {
      "epoch": 0.07705615223816045,
      "grad_norm": 0.13727741163672458,
      "learning_rate": 0.00019884363696398636,
      "loss": 1.0404,
      "step": 1069
    },
    {
      "epoch": 0.07712823470049737,
      "grad_norm": 0.7945065858620427,
      "learning_rate": 0.0001988400939963482,
      "loss": 1.1538,
      "step": 1070
    },
    {
      "epoch": 0.07720031716283428,
      "grad_norm": 0.13584837980042785,
      "learning_rate": 0.00019883654564104068,
      "loss": 1.1689,
      "step": 1071
    },
    {
      "epoch": 0.0772723996251712,
      "grad_norm": 0.5846826876967549,
      "learning_rate": 0.00019883299189825724,
      "loss": 1.1914,
      "step": 1072
    },
    {
      "epoch": 0.0773444820875081,
      "grad_norm": 0.9368065662934947,
      "learning_rate": 0.00019882943276819154,
      "loss": 1.1804,
      "step": 1073
    },
    {
      "epoch": 0.07741656454984502,
      "grad_norm": 0.19046477223167105,
      "learning_rate": 0.00019882586825103758,
      "loss": 1.0714,
      "step": 1074
    },
    {
      "epoch": 0.07748864701218193,
      "grad_norm": 0.19746332197571453,
      "learning_rate": 0.00019882229834698967,
      "loss": 1.0484,
      "step": 1075
    },
    {
      "epoch": 0.07756072947451885,
      "grad_norm": 0.28866163536474654,
      "learning_rate": 0.00019881872305624242,
      "loss": 1.0935,
      "step": 1076
    },
    {
      "epoch": 0.07763281193685577,
      "grad_norm": 0.2045416382188679,
      "learning_rate": 0.00019881514237899073,
      "loss": 1.1851,
      "step": 1077
    },
    {
      "epoch": 0.07770489439919268,
      "grad_norm": 0.680707744955549,
      "learning_rate": 0.00019881155631542975,
      "loss": 1.1134,
      "step": 1078
    },
    {
      "epoch": 0.0777769768615296,
      "grad_norm": 1.9331548154404863,
      "learning_rate": 0.00019880796486575496,
      "loss": 1.2057,
      "step": 1079
    },
    {
      "epoch": 0.0778490593238665,
      "grad_norm": 1.3940488497939265,
      "learning_rate": 0.00019880436803016208,
      "loss": 1.0888,
      "step": 1080
    },
    {
      "epoch": 0.07792114178620342,
      "grad_norm": 0.9269738965014735,
      "learning_rate": 0.00019880076580884722,
      "loss": 1.2218,
      "step": 1081
    },
    {
      "epoch": 0.07799322424854033,
      "grad_norm": 0.13633264533071654,
      "learning_rate": 0.00019879715820200677,
      "loss": 1.1801,
      "step": 1082
    },
    {
      "epoch": 0.07806530671087725,
      "grad_norm": 0.3967308818705607,
      "learning_rate": 0.00019879354520983733,
      "loss": 1.0881,
      "step": 1083
    },
    {
      "epoch": 0.07813738917321415,
      "grad_norm": 0.24153410156204438,
      "learning_rate": 0.00019878992683253582,
      "loss": 1.1638,
      "step": 1084
    },
    {
      "epoch": 0.07820947163555107,
      "grad_norm": 0.41828319435205236,
      "learning_rate": 0.0001987863030702995,
      "loss": 1.0643,
      "step": 1085
    },
    {
      "epoch": 0.07828155409788798,
      "grad_norm": 0.12625822781937043,
      "learning_rate": 0.0001987826739233259,
      "loss": 1.1457,
      "step": 1086
    },
    {
      "epoch": 0.0783536365602249,
      "grad_norm": 0.12904809298011638,
      "learning_rate": 0.00019877903939181284,
      "loss": 1.0765,
      "step": 1087
    },
    {
      "epoch": 0.0784257190225618,
      "grad_norm": 0.33269107497845746,
      "learning_rate": 0.00019877539947595844,
      "loss": 1.1107,
      "step": 1088
    },
    {
      "epoch": 0.07849780148489872,
      "grad_norm": 0.16560964944195997,
      "learning_rate": 0.00019877175417596106,
      "loss": 1.1483,
      "step": 1089
    },
    {
      "epoch": 0.07856988394723564,
      "grad_norm": 0.513618170437866,
      "learning_rate": 0.00019876810349201947,
      "loss": 1.1804,
      "step": 1090
    },
    {
      "epoch": 0.07864196640957255,
      "grad_norm": 0.20018769148269383,
      "learning_rate": 0.00019876444742433265,
      "loss": 1.0761,
      "step": 1091
    },
    {
      "epoch": 0.07871404887190947,
      "grad_norm": 0.4244984662431924,
      "learning_rate": 0.00019876078597309984,
      "loss": 1.1041,
      "step": 1092
    },
    {
      "epoch": 0.07878613133424638,
      "grad_norm": 0.24526502244819548,
      "learning_rate": 0.00019875711913852067,
      "loss": 1.1381,
      "step": 1093
    },
    {
      "epoch": 0.0788582137965833,
      "grad_norm": 0.16064922491899405,
      "learning_rate": 0.000198753446920795,
      "loss": 1.2513,
      "step": 1094
    },
    {
      "epoch": 0.0789302962589202,
      "grad_norm": 0.19039355124208093,
      "learning_rate": 0.000198749769320123,
      "loss": 1.1171,
      "step": 1095
    },
    {
      "epoch": 0.07900237872125712,
      "grad_norm": 0.15233007553576994,
      "learning_rate": 0.00019874608633670512,
      "loss": 1.0656,
      "step": 1096
    },
    {
      "epoch": 0.07907446118359403,
      "grad_norm": 0.2117091510576692,
      "learning_rate": 0.00019874239797074212,
      "loss": 1.1657,
      "step": 1097
    },
    {
      "epoch": 0.07914654364593095,
      "grad_norm": 1.1388445085769932,
      "learning_rate": 0.00019873870422243507,
      "loss": 1.1566,
      "step": 1098
    },
    {
      "epoch": 0.07921862610826785,
      "grad_norm": 0.13464742082625517,
      "learning_rate": 0.0001987350050919853,
      "loss": 1.1487,
      "step": 1099
    },
    {
      "epoch": 0.07929070857060477,
      "grad_norm": 0.9129055730244041,
      "learning_rate": 0.00019873130057959445,
      "loss": 1.1492,
      "step": 1100
    },
    {
      "epoch": 0.07936279103294168,
      "grad_norm": 0.18839408608822186,
      "learning_rate": 0.00019872759068546446,
      "loss": 1.1556,
      "step": 1101
    },
    {
      "epoch": 0.0794348734952786,
      "grad_norm": 0.21572349055714457,
      "learning_rate": 0.00019872387540979748,
      "loss": 1.1258,
      "step": 1102
    },
    {
      "epoch": 0.07950695595761552,
      "grad_norm": 0.7358412629471237,
      "learning_rate": 0.00019872015475279616,
      "loss": 1.0974,
      "step": 1103
    },
    {
      "epoch": 0.07957903841995242,
      "grad_norm": 0.20109776950007438,
      "learning_rate": 0.00019871642871466315,
      "loss": 1.1681,
      "step": 1104
    },
    {
      "epoch": 0.07965112088228934,
      "grad_norm": 0.11514712000160329,
      "learning_rate": 0.00019871269729560166,
      "loss": 1.13,
      "step": 1105
    },
    {
      "epoch": 0.07972320334462625,
      "grad_norm": 0.3356143412557696,
      "learning_rate": 0.0001987089604958151,
      "loss": 1.1823,
      "step": 1106
    },
    {
      "epoch": 0.07979528580696317,
      "grad_norm": 0.43347106689714143,
      "learning_rate": 0.00019870521831550706,
      "loss": 1.1947,
      "step": 1107
    },
    {
      "epoch": 0.07986736826930008,
      "grad_norm": 0.14563849578315874,
      "learning_rate": 0.0001987014707548816,
      "loss": 1.0321,
      "step": 1108
    },
    {
      "epoch": 0.079939450731637,
      "grad_norm": 0.2517719561126084,
      "learning_rate": 0.000198697717814143,
      "loss": 1.0867,
      "step": 1109
    },
    {
      "epoch": 0.0800115331939739,
      "grad_norm": 0.11177750347973833,
      "learning_rate": 0.00019869395949349578,
      "loss": 1.1861,
      "step": 1110
    },
    {
      "epoch": 0.08008361565631082,
      "grad_norm": 0.22537121517989644,
      "learning_rate": 0.0001986901957931448,
      "loss": 1.0616,
      "step": 1111
    },
    {
      "epoch": 0.08015569811864773,
      "grad_norm": 0.5078300236598522,
      "learning_rate": 0.0001986864267132953,
      "loss": 1.2299,
      "step": 1112
    },
    {
      "epoch": 0.08022778058098465,
      "grad_norm": 0.1467623082944967,
      "learning_rate": 0.00019868265225415265,
      "loss": 1.0447,
      "step": 1113
    },
    {
      "epoch": 0.08029986304332155,
      "grad_norm": 0.13497297724921575,
      "learning_rate": 0.00019867887241592258,
      "loss": 1.1112,
      "step": 1114
    },
    {
      "epoch": 0.08037194550565847,
      "grad_norm": 0.1402031868991692,
      "learning_rate": 0.00019867508719881118,
      "loss": 1.0948,
      "step": 1115
    },
    {
      "epoch": 0.0804440279679954,
      "grad_norm": 0.2610417751174269,
      "learning_rate": 0.00019867129660302476,
      "loss": 1.1408,
      "step": 1116
    },
    {
      "epoch": 0.0805161104303323,
      "grad_norm": 0.33919343757642034,
      "learning_rate": 0.00019866750062876994,
      "loss": 1.0505,
      "step": 1117
    },
    {
      "epoch": 0.08058819289266922,
      "grad_norm": 0.11842251373672927,
      "learning_rate": 0.00019866369927625362,
      "loss": 1.0575,
      "step": 1118
    },
    {
      "epoch": 0.08066027535500613,
      "grad_norm": 0.40766443480974157,
      "learning_rate": 0.000198659892545683,
      "loss": 1.1711,
      "step": 1119
    },
    {
      "epoch": 0.08073235781734305,
      "grad_norm": 0.25061677805917537,
      "learning_rate": 0.0001986560804372656,
      "loss": 1.0835,
      "step": 1120
    },
    {
      "epoch": 0.08080444027967995,
      "grad_norm": 1.1871044646135844,
      "learning_rate": 0.00019865226295120926,
      "loss": 1.0466,
      "step": 1121
    },
    {
      "epoch": 0.08087652274201687,
      "grad_norm": 0.32266840387019435,
      "learning_rate": 0.000198648440087722,
      "loss": 1.2003,
      "step": 1122
    },
    {
      "epoch": 0.08094860520435378,
      "grad_norm": 2.433208357763416,
      "learning_rate": 0.0001986446118470122,
      "loss": 1.1615,
      "step": 1123
    },
    {
      "epoch": 0.0810206876666907,
      "grad_norm": 0.11743474124060181,
      "learning_rate": 0.00019864077822928858,
      "loss": 1.1476,
      "step": 1124
    },
    {
      "epoch": 0.0810927701290276,
      "grad_norm": 0.23293521340522053,
      "learning_rate": 0.00019863693923476006,
      "loss": 1.0612,
      "step": 1125
    },
    {
      "epoch": 0.08116485259136452,
      "grad_norm": 0.20516260628047447,
      "learning_rate": 0.00019863309486363592,
      "loss": 1.087,
      "step": 1126
    },
    {
      "epoch": 0.08123693505370143,
      "grad_norm": 0.1487805909855085,
      "learning_rate": 0.00019862924511612575,
      "loss": 1.1259,
      "step": 1127
    },
    {
      "epoch": 0.08130901751603835,
      "grad_norm": 0.14262124816523605,
      "learning_rate": 0.00019862538999243932,
      "loss": 1.1506,
      "step": 1128
    },
    {
      "epoch": 0.08138109997837527,
      "grad_norm": 0.20634560359918638,
      "learning_rate": 0.0001986215294927868,
      "loss": 1.2182,
      "step": 1129
    },
    {
      "epoch": 0.08145318244071217,
      "grad_norm": 0.20973296762932678,
      "learning_rate": 0.00019861766361737866,
      "loss": 1.0704,
      "step": 1130
    },
    {
      "epoch": 0.0815252649030491,
      "grad_norm": 0.1918294611278691,
      "learning_rate": 0.00019861379236642558,
      "loss": 1.1359,
      "step": 1131
    },
    {
      "epoch": 0.081597347365386,
      "grad_norm": 0.4033174922672694,
      "learning_rate": 0.00019860991574013862,
      "loss": 1.1067,
      "step": 1132
    },
    {
      "epoch": 0.08166942982772292,
      "grad_norm": 0.3028458787965476,
      "learning_rate": 0.00019860603373872903,
      "loss": 1.1042,
      "step": 1133
    },
    {
      "epoch": 0.08174151229005983,
      "grad_norm": 0.14616571314295942,
      "learning_rate": 0.00019860214636240845,
      "loss": 1.0929,
      "step": 1134
    },
    {
      "epoch": 0.08181359475239675,
      "grad_norm": 0.23775635073789286,
      "learning_rate": 0.00019859825361138878,
      "loss": 1.0886,
      "step": 1135
    },
    {
      "epoch": 0.08188567721473365,
      "grad_norm": 0.1961817203579509,
      "learning_rate": 0.00019859435548588218,
      "loss": 1.157,
      "step": 1136
    },
    {
      "epoch": 0.08195775967707057,
      "grad_norm": 1.2205053817585882,
      "learning_rate": 0.00019859045198610117,
      "loss": 1.054,
      "step": 1137
    },
    {
      "epoch": 0.08202984213940748,
      "grad_norm": 0.2315629692280198,
      "learning_rate": 0.00019858654311225853,
      "loss": 1.1612,
      "step": 1138
    },
    {
      "epoch": 0.0821019246017444,
      "grad_norm": 0.7829621582242292,
      "learning_rate": 0.00019858262886456732,
      "loss": 1.1232,
      "step": 1139
    },
    {
      "epoch": 0.0821740070640813,
      "grad_norm": 0.1327665571568982,
      "learning_rate": 0.00019857870924324086,
      "loss": 1.1518,
      "step": 1140
    },
    {
      "epoch": 0.08224608952641822,
      "grad_norm": 0.12167082859405519,
      "learning_rate": 0.00019857478424849286,
      "loss": 1.0951,
      "step": 1141
    },
    {
      "epoch": 0.08231817198875514,
      "grad_norm": 0.1328681062709836,
      "learning_rate": 0.00019857085388053723,
      "loss": 1.1504,
      "step": 1142
    },
    {
      "epoch": 0.08239025445109205,
      "grad_norm": 0.29904343984869286,
      "learning_rate": 0.0001985669181395882,
      "loss": 1.1178,
      "step": 1143
    },
    {
      "epoch": 0.08246233691342897,
      "grad_norm": 0.13502270695578383,
      "learning_rate": 0.00019856297702586036,
      "loss": 1.1208,
      "step": 1144
    },
    {
      "epoch": 0.08253441937576587,
      "grad_norm": 0.2146690649374417,
      "learning_rate": 0.00019855903053956847,
      "loss": 1.1819,
      "step": 1145
    },
    {
      "epoch": 0.0826065018381028,
      "grad_norm": 0.9868911737043962,
      "learning_rate": 0.0001985550786809277,
      "loss": 1.1287,
      "step": 1146
    },
    {
      "epoch": 0.0826785843004397,
      "grad_norm": 0.14508540357645366,
      "learning_rate": 0.00019855112145015347,
      "loss": 1.1263,
      "step": 1147
    },
    {
      "epoch": 0.08275066676277662,
      "grad_norm": 0.13123691625255196,
      "learning_rate": 0.0001985471588474614,
      "loss": 1.1239,
      "step": 1148
    },
    {
      "epoch": 0.08282274922511353,
      "grad_norm": 0.1691318565741071,
      "learning_rate": 0.0001985431908730676,
      "loss": 1.0642,
      "step": 1149
    },
    {
      "epoch": 0.08289483168745045,
      "grad_norm": 0.11696044737530584,
      "learning_rate": 0.00019853921752718828,
      "loss": 1.1521,
      "step": 1150
    },
    {
      "epoch": 0.08296691414978735,
      "grad_norm": 0.13584264395664636,
      "learning_rate": 0.00019853523881004006,
      "loss": 1.0557,
      "step": 1151
    },
    {
      "epoch": 0.08303899661212427,
      "grad_norm": 0.485702244117622,
      "learning_rate": 0.00019853125472183979,
      "loss": 1.1431,
      "step": 1152
    },
    {
      "epoch": 0.08311107907446118,
      "grad_norm": 0.18141694218619953,
      "learning_rate": 0.00019852726526280466,
      "loss": 1.186,
      "step": 1153
    },
    {
      "epoch": 0.0831831615367981,
      "grad_norm": 0.130091535933995,
      "learning_rate": 0.00019852327043315212,
      "loss": 1.1145,
      "step": 1154
    },
    {
      "epoch": 0.08325524399913502,
      "grad_norm": 0.19207363790344076,
      "learning_rate": 0.0001985192702330999,
      "loss": 1.2036,
      "step": 1155
    },
    {
      "epoch": 0.08332732646147192,
      "grad_norm": 0.1275329329523737,
      "learning_rate": 0.0001985152646628661,
      "loss": 1.1203,
      "step": 1156
    },
    {
      "epoch": 0.08339940892380884,
      "grad_norm": 0.18346175063045514,
      "learning_rate": 0.00019851125372266902,
      "loss": 1.1068,
      "step": 1157
    },
    {
      "epoch": 0.08347149138614575,
      "grad_norm": 2.525707509541452,
      "learning_rate": 0.00019850723741272734,
      "loss": 1.0931,
      "step": 1158
    },
    {
      "epoch": 0.08354357384848267,
      "grad_norm": 0.21065681520534396,
      "learning_rate": 0.00019850321573325992,
      "loss": 1.1013,
      "step": 1159
    },
    {
      "epoch": 0.08361565631081957,
      "grad_norm": 0.1356981643569954,
      "learning_rate": 0.00019849918868448602,
      "loss": 1.0557,
      "step": 1160
    },
    {
      "epoch": 0.0836877387731565,
      "grad_norm": 0.1224681776535988,
      "learning_rate": 0.0001984951562666251,
      "loss": 1.1181,
      "step": 1161
    },
    {
      "epoch": 0.0837598212354934,
      "grad_norm": 0.25584502759661754,
      "learning_rate": 0.00019849111847989706,
      "loss": 1.0905,
      "step": 1162
    },
    {
      "epoch": 0.08383190369783032,
      "grad_norm": 0.18680376722307207,
      "learning_rate": 0.00019848707532452191,
      "loss": 1.1152,
      "step": 1163
    },
    {
      "epoch": 0.08390398616016723,
      "grad_norm": 0.26021441979097837,
      "learning_rate": 0.00019848302680072007,
      "loss": 1.0765,
      "step": 1164
    },
    {
      "epoch": 0.08397606862250415,
      "grad_norm": 0.14212215464726974,
      "learning_rate": 0.0001984789729087122,
      "loss": 1.0145,
      "step": 1165
    },
    {
      "epoch": 0.08404815108484105,
      "grad_norm": 0.19491077902646353,
      "learning_rate": 0.00019847491364871931,
      "loss": 1.1417,
      "step": 1166
    },
    {
      "epoch": 0.08412023354717797,
      "grad_norm": 0.22178817520841143,
      "learning_rate": 0.00019847084902096263,
      "loss": 1.1326,
      "step": 1167
    },
    {
      "epoch": 0.08419231600951488,
      "grad_norm": 0.2269957314882568,
      "learning_rate": 0.0001984667790256637,
      "loss": 1.1308,
      "step": 1168
    },
    {
      "epoch": 0.0842643984718518,
      "grad_norm": 0.15103195544048664,
      "learning_rate": 0.00019846270366304447,
      "loss": 1.0796,
      "step": 1169
    },
    {
      "epoch": 0.08433648093418872,
      "grad_norm": 0.23802121361894726,
      "learning_rate": 0.00019845862293332698,
      "loss": 1.1636,
      "step": 1170
    },
    {
      "epoch": 0.08440856339652562,
      "grad_norm": 0.12375378050226472,
      "learning_rate": 0.00019845453683673368,
      "loss": 1.0675,
      "step": 1171
    },
    {
      "epoch": 0.08448064585886254,
      "grad_norm": 0.14708409599399278,
      "learning_rate": 0.00019845044537348738,
      "loss": 1.0283,
      "step": 1172
    },
    {
      "epoch": 0.08455272832119945,
      "grad_norm": 0.13653680039236432,
      "learning_rate": 0.00019844634854381102,
      "loss": 1.0606,
      "step": 1173
    },
    {
      "epoch": 0.08462481078353637,
      "grad_norm": 0.1571616064965639,
      "learning_rate": 0.00019844224634792793,
      "loss": 1.0624,
      "step": 1174
    },
    {
      "epoch": 0.08469689324587328,
      "grad_norm": 0.5411236034591238,
      "learning_rate": 0.00019843813878606174,
      "loss": 1.1005,
      "step": 1175
    },
    {
      "epoch": 0.0847689757082102,
      "grad_norm": 0.13589776297258607,
      "learning_rate": 0.00019843402585843631,
      "loss": 1.1606,
      "step": 1176
    },
    {
      "epoch": 0.0848410581705471,
      "grad_norm": 0.13021334091498724,
      "learning_rate": 0.00019842990756527588,
      "loss": 1.1457,
      "step": 1177
    },
    {
      "epoch": 0.08491314063288402,
      "grad_norm": 0.5494567179897721,
      "learning_rate": 0.00019842578390680488,
      "loss": 1.084,
      "step": 1178
    },
    {
      "epoch": 0.08498522309522093,
      "grad_norm": 0.12644422748320513,
      "learning_rate": 0.00019842165488324814,
      "loss": 1.0786,
      "step": 1179
    },
    {
      "epoch": 0.08505730555755785,
      "grad_norm": 0.12792742481957245,
      "learning_rate": 0.0001984175204948307,
      "loss": 1.148,
      "step": 1180
    },
    {
      "epoch": 0.08512938801989475,
      "grad_norm": 0.1709761905512325,
      "learning_rate": 0.00019841338074177793,
      "loss": 1.1592,
      "step": 1181
    },
    {
      "epoch": 0.08520147048223167,
      "grad_norm": 0.16938680097606326,
      "learning_rate": 0.00019840923562431545,
      "loss": 1.1545,
      "step": 1182
    },
    {
      "epoch": 0.08527355294456859,
      "grad_norm": 0.5343639144529624,
      "learning_rate": 0.00019840508514266928,
      "loss": 1.0564,
      "step": 1183
    },
    {
      "epoch": 0.0853456354069055,
      "grad_norm": 0.15267359149334886,
      "learning_rate": 0.00019840092929706557,
      "loss": 1.0989,
      "step": 1184
    },
    {
      "epoch": 0.08541771786924242,
      "grad_norm": 0.3024668034106112,
      "learning_rate": 0.0001983967680877309,
      "loss": 1.1731,
      "step": 1185
    },
    {
      "epoch": 0.08548980033157932,
      "grad_norm": 0.1307710448484827,
      "learning_rate": 0.00019839260151489213,
      "loss": 1.0843,
      "step": 1186
    },
    {
      "epoch": 0.08556188279391624,
      "grad_norm": 0.119853167113074,
      "learning_rate": 0.00019838842957877625,
      "loss": 1.1419,
      "step": 1187
    },
    {
      "epoch": 0.08563396525625315,
      "grad_norm": 0.13277992990305965,
      "learning_rate": 0.00019838425227961082,
      "loss": 1.0695,
      "step": 1188
    },
    {
      "epoch": 0.08570604771859007,
      "grad_norm": 0.19786804709722905,
      "learning_rate": 0.00019838006961762344,
      "loss": 1.1025,
      "step": 1189
    },
    {
      "epoch": 0.08577813018092698,
      "grad_norm": 0.12947877665043914,
      "learning_rate": 0.00019837588159304215,
      "loss": 1.1546,
      "step": 1190
    },
    {
      "epoch": 0.0858502126432639,
      "grad_norm": 0.5095625211989246,
      "learning_rate": 0.00019837168820609522,
      "loss": 1.0841,
      "step": 1191
    },
    {
      "epoch": 0.0859222951056008,
      "grad_norm": 0.13434271314359927,
      "learning_rate": 0.0001983674894570112,
      "loss": 1.1686,
      "step": 1192
    },
    {
      "epoch": 0.08599437756793772,
      "grad_norm": 0.1797196587330627,
      "learning_rate": 0.000198363285346019,
      "loss": 1.1219,
      "step": 1193
    },
    {
      "epoch": 0.08606646003027463,
      "grad_norm": 0.12309192812612779,
      "learning_rate": 0.00019835907587334775,
      "loss": 1.1403,
      "step": 1194
    },
    {
      "epoch": 0.08613854249261155,
      "grad_norm": 0.12157241203035828,
      "learning_rate": 0.00019835486103922693,
      "loss": 1.169,
      "step": 1195
    },
    {
      "epoch": 0.08621062495494847,
      "grad_norm": 0.1118627524513195,
      "learning_rate": 0.00019835064084388628,
      "loss": 1.0957,
      "step": 1196
    },
    {
      "epoch": 0.08628270741728537,
      "grad_norm": 0.16566670209270484,
      "learning_rate": 0.00019834641528755582,
      "loss": 1.2125,
      "step": 1197
    },
    {
      "epoch": 0.08635478987962229,
      "grad_norm": 0.896953919600734,
      "learning_rate": 0.0001983421843704659,
      "loss": 1.0843,
      "step": 1198
    },
    {
      "epoch": 0.0864268723419592,
      "grad_norm": 0.2214999794574688,
      "learning_rate": 0.00019833794809284716,
      "loss": 1.1235,
      "step": 1199
    },
    {
      "epoch": 0.08649895480429612,
      "grad_norm": 0.12860249707297136,
      "learning_rate": 0.00019833370645493047,
      "loss": 1.072,
      "step": 1200
    },
    {
      "epoch": 0.08657103726663302,
      "grad_norm": 0.14121035588737835,
      "learning_rate": 0.00019832945945694706,
      "loss": 1.1182,
      "step": 1201
    },
    {
      "epoch": 0.08664311972896994,
      "grad_norm": 0.11512471922270617,
      "learning_rate": 0.00019832520709912844,
      "loss": 1.1065,
      "step": 1202
    },
    {
      "epoch": 0.08671520219130685,
      "grad_norm": 0.16265838104723176,
      "learning_rate": 0.00019832094938170637,
      "loss": 1.1205,
      "step": 1203
    },
    {
      "epoch": 0.08678728465364377,
      "grad_norm": 0.2721976012350073,
      "learning_rate": 0.00019831668630491297,
      "loss": 1.1364,
      "step": 1204
    },
    {
      "epoch": 0.08685936711598068,
      "grad_norm": 0.2483622923145994,
      "learning_rate": 0.00019831241786898062,
      "loss": 1.1848,
      "step": 1205
    },
    {
      "epoch": 0.0869314495783176,
      "grad_norm": 0.13245273697572607,
      "learning_rate": 0.00019830814407414193,
      "loss": 1.1098,
      "step": 1206
    },
    {
      "epoch": 0.0870035320406545,
      "grad_norm": 0.28401818365852793,
      "learning_rate": 0.00019830386492062992,
      "loss": 1.1105,
      "step": 1207
    },
    {
      "epoch": 0.08707561450299142,
      "grad_norm": 0.1338621946640475,
      "learning_rate": 0.00019829958040867783,
      "loss": 1.0677,
      "step": 1208
    },
    {
      "epoch": 0.08714769696532834,
      "grad_norm": 0.1290899901006582,
      "learning_rate": 0.00019829529053851919,
      "loss": 1.1035,
      "step": 1209
    },
    {
      "epoch": 0.08721977942766525,
      "grad_norm": 0.18965103503625222,
      "learning_rate": 0.00019829099531038782,
      "loss": 1.0973,
      "step": 1210
    },
    {
      "epoch": 0.08729186189000217,
      "grad_norm": 0.23034439076472762,
      "learning_rate": 0.0001982866947245179,
      "loss": 1.1395,
      "step": 1211
    },
    {
      "epoch": 0.08736394435233907,
      "grad_norm": 0.2665142752504326,
      "learning_rate": 0.0001982823887811438,
      "loss": 1.141,
      "step": 1212
    },
    {
      "epoch": 0.087436026814676,
      "grad_norm": 0.40400781030838284,
      "learning_rate": 0.00019827807748050023,
      "loss": 1.1653,
      "step": 1213
    },
    {
      "epoch": 0.0875081092770129,
      "grad_norm": 0.14303011925968792,
      "learning_rate": 0.00019827376082282223,
      "loss": 1.1667,
      "step": 1214
    },
    {
      "epoch": 0.08758019173934982,
      "grad_norm": 0.23291333947940335,
      "learning_rate": 0.0001982694388083451,
      "loss": 1.0615,
      "step": 1215
    },
    {
      "epoch": 0.08765227420168673,
      "grad_norm": 0.1729318965039834,
      "learning_rate": 0.0001982651114373044,
      "loss": 1.1757,
      "step": 1216
    },
    {
      "epoch": 0.08772435666402364,
      "grad_norm": 0.1922243431626503,
      "learning_rate": 0.000198260778709936,
      "loss": 1.1783,
      "step": 1217
    },
    {
      "epoch": 0.08779643912636055,
      "grad_norm": 0.13879670115924958,
      "learning_rate": 0.00019825644062647615,
      "loss": 1.1308,
      "step": 1218
    },
    {
      "epoch": 0.08786852158869747,
      "grad_norm": 0.1460264776156725,
      "learning_rate": 0.00019825209718716123,
      "loss": 1.1744,
      "step": 1219
    },
    {
      "epoch": 0.08794060405103438,
      "grad_norm": 0.11965048091649028,
      "learning_rate": 0.00019824774839222802,
      "loss": 1.1418,
      "step": 1220
    },
    {
      "epoch": 0.0880126865133713,
      "grad_norm": 0.1186474937144144,
      "learning_rate": 0.00019824339424191355,
      "loss": 1.1088,
      "step": 1221
    },
    {
      "epoch": 0.08808476897570822,
      "grad_norm": 0.11053703846913866,
      "learning_rate": 0.00019823903473645523,
      "loss": 1.1133,
      "step": 1222
    },
    {
      "epoch": 0.08815685143804512,
      "grad_norm": 0.13951661678343558,
      "learning_rate": 0.00019823466987609062,
      "loss": 1.116,
      "step": 1223
    },
    {
      "epoch": 0.08822893390038204,
      "grad_norm": 0.4608134074085216,
      "learning_rate": 0.0001982302996610577,
      "loss": 1.1275,
      "step": 1224
    },
    {
      "epoch": 0.08830101636271895,
      "grad_norm": 0.10971282080065407,
      "learning_rate": 0.00019822592409159462,
      "loss": 1.11,
      "step": 1225
    },
    {
      "epoch": 0.08837309882505587,
      "grad_norm": 0.45950611108268014,
      "learning_rate": 0.00019822154316793993,
      "loss": 1.1523,
      "step": 1226
    },
    {
      "epoch": 0.08844518128739277,
      "grad_norm": 0.234861778353743,
      "learning_rate": 0.00019821715689033242,
      "loss": 1.1656,
      "step": 1227
    },
    {
      "epoch": 0.0885172637497297,
      "grad_norm": 0.14786225325086838,
      "learning_rate": 0.0001982127652590112,
      "loss": 1.0903,
      "step": 1228
    },
    {
      "epoch": 0.0885893462120666,
      "grad_norm": 0.1499896105970178,
      "learning_rate": 0.0001982083682742156,
      "loss": 1.193,
      "step": 1229
    },
    {
      "epoch": 0.08866142867440352,
      "grad_norm": 0.15430019953933616,
      "learning_rate": 0.00019820396593618535,
      "loss": 1.0644,
      "step": 1230
    },
    {
      "epoch": 0.08873351113674043,
      "grad_norm": 0.1444498176869029,
      "learning_rate": 0.00019819955824516044,
      "loss": 1.0702,
      "step": 1231
    },
    {
      "epoch": 0.08880559359907735,
      "grad_norm": 0.13530633176224377,
      "learning_rate": 0.00019819514520138105,
      "loss": 1.11,
      "step": 1232
    },
    {
      "epoch": 0.08887767606141425,
      "grad_norm": 0.132732418820427,
      "learning_rate": 0.00019819072680508776,
      "loss": 1.1173,
      "step": 1233
    },
    {
      "epoch": 0.08894975852375117,
      "grad_norm": 0.20690754146442292,
      "learning_rate": 0.0001981863030565214,
      "loss": 1.1978,
      "step": 1234
    },
    {
      "epoch": 0.08902184098608809,
      "grad_norm": 0.11946463916045834,
      "learning_rate": 0.0001981818739559232,
      "loss": 1.0632,
      "step": 1235
    },
    {
      "epoch": 0.089093923448425,
      "grad_norm": 0.16584847862353663,
      "learning_rate": 0.00019817743950353443,
      "loss": 1.1264,
      "step": 1236
    },
    {
      "epoch": 0.08916600591076192,
      "grad_norm": 0.12274716732548926,
      "learning_rate": 0.00019817299969959693,
      "loss": 1.0874,
      "step": 1237
    },
    {
      "epoch": 0.08923808837309882,
      "grad_norm": 0.1715283022461144,
      "learning_rate": 0.00019816855454435266,
      "loss": 1.1275,
      "step": 1238
    },
    {
      "epoch": 0.08931017083543574,
      "grad_norm": 0.11497728917777256,
      "learning_rate": 0.00019816410403804394,
      "loss": 1.0774,
      "step": 1239
    },
    {
      "epoch": 0.08938225329777265,
      "grad_norm": 0.13313646105482801,
      "learning_rate": 0.0001981596481809133,
      "loss": 1.1861,
      "step": 1240
    },
    {
      "epoch": 0.08945433576010957,
      "grad_norm": 0.15627951986214794,
      "learning_rate": 0.00019815518697320373,
      "loss": 1.0828,
      "step": 1241
    },
    {
      "epoch": 0.08952641822244647,
      "grad_norm": 0.2970601787696286,
      "learning_rate": 0.0001981507204151583,
      "loss": 1.197,
      "step": 1242
    },
    {
      "epoch": 0.0895985006847834,
      "grad_norm": 0.13322056028006535,
      "learning_rate": 0.00019814624850702055,
      "loss": 1.1027,
      "step": 1243
    },
    {
      "epoch": 0.0896705831471203,
      "grad_norm": 1.490998423228474,
      "learning_rate": 0.00019814177124903421,
      "loss": 1.079,
      "step": 1244
    },
    {
      "epoch": 0.08974266560945722,
      "grad_norm": 0.1827676012826355,
      "learning_rate": 0.00019813728864144335,
      "loss": 1.0721,
      "step": 1245
    },
    {
      "epoch": 0.08981474807179413,
      "grad_norm": 0.14163961428639074,
      "learning_rate": 0.00019813280068449227,
      "loss": 1.1663,
      "step": 1246
    },
    {
      "epoch": 0.08988683053413105,
      "grad_norm": 0.19951952353562563,
      "learning_rate": 0.00019812830737842566,
      "loss": 1.2004,
      "step": 1247
    },
    {
      "epoch": 0.08995891299646797,
      "grad_norm": 0.12488452325489074,
      "learning_rate": 0.0001981238087234884,
      "loss": 1.1083,
      "step": 1248
    },
    {
      "epoch": 0.09003099545880487,
      "grad_norm": 0.1840273324995988,
      "learning_rate": 0.00019811930471992572,
      "loss": 1.1029,
      "step": 1249
    },
    {
      "epoch": 0.09010307792114179,
      "grad_norm": 0.11800294783191721,
      "learning_rate": 0.00019811479536798314,
      "loss": 1.1515,
      "step": 1250
    },
    {
      "epoch": 0.0901751603834787,
      "grad_norm": 0.11480520350672906,
      "learning_rate": 0.00019811028066790644,
      "loss": 1.1405,
      "step": 1251
    },
    {
      "epoch": 0.09024724284581562,
      "grad_norm": 0.15417845123856652,
      "learning_rate": 0.00019810576061994177,
      "loss": 1.1129,
      "step": 1252
    },
    {
      "epoch": 0.09031932530815252,
      "grad_norm": 0.12457936461502507,
      "learning_rate": 0.00019810123522433543,
      "loss": 1.0512,
      "step": 1253
    },
    {
      "epoch": 0.09039140777048944,
      "grad_norm": 0.14528980396054159,
      "learning_rate": 0.00019809670448133412,
      "loss": 1.105,
      "step": 1254
    },
    {
      "epoch": 0.09046349023282635,
      "grad_norm": 0.19901088054772428,
      "learning_rate": 0.00019809216839118484,
      "loss": 1.1222,
      "step": 1255
    },
    {
      "epoch": 0.09053557269516327,
      "grad_norm": 0.1283578863140042,
      "learning_rate": 0.00019808762695413482,
      "loss": 1.1286,
      "step": 1256
    },
    {
      "epoch": 0.09060765515750017,
      "grad_norm": 0.12028920469433668,
      "learning_rate": 0.0001980830801704316,
      "loss": 1.1358,
      "step": 1257
    },
    {
      "epoch": 0.0906797376198371,
      "grad_norm": 0.3707033566258525,
      "learning_rate": 0.00019807852804032305,
      "loss": 1.0382,
      "step": 1258
    },
    {
      "epoch": 0.090751820082174,
      "grad_norm": 0.2644320482196782,
      "learning_rate": 0.0001980739705640573,
      "loss": 1.2217,
      "step": 1259
    },
    {
      "epoch": 0.09082390254451092,
      "grad_norm": 0.14418535669040794,
      "learning_rate": 0.00019806940774188273,
      "loss": 1.0903,
      "step": 1260
    },
    {
      "epoch": 0.09089598500684784,
      "grad_norm": 0.14166159313350593,
      "learning_rate": 0.00019806483957404808,
      "loss": 1.0594,
      "step": 1261
    },
    {
      "epoch": 0.09096806746918475,
      "grad_norm": 0.21692168337679593,
      "learning_rate": 0.0001980602660608024,
      "loss": 1.1344,
      "step": 1262
    },
    {
      "epoch": 0.09104014993152167,
      "grad_norm": 0.2123324369390397,
      "learning_rate": 0.0001980556872023949,
      "loss": 1.0753,
      "step": 1263
    },
    {
      "epoch": 0.09111223239385857,
      "grad_norm": 0.14579857048396777,
      "learning_rate": 0.00019805110299907526,
      "loss": 1.1607,
      "step": 1264
    },
    {
      "epoch": 0.09118431485619549,
      "grad_norm": 0.1543471727346113,
      "learning_rate": 0.00019804651345109328,
      "loss": 1.0791,
      "step": 1265
    },
    {
      "epoch": 0.0912563973185324,
      "grad_norm": 0.1794602808162132,
      "learning_rate": 0.0001980419185586992,
      "loss": 1.0629,
      "step": 1266
    },
    {
      "epoch": 0.09132847978086932,
      "grad_norm": 0.1472286418223704,
      "learning_rate": 0.00019803731832214344,
      "loss": 1.1725,
      "step": 1267
    },
    {
      "epoch": 0.09140056224320622,
      "grad_norm": 0.11383887067533179,
      "learning_rate": 0.00019803271274167674,
      "loss": 1.0882,
      "step": 1268
    },
    {
      "epoch": 0.09147264470554314,
      "grad_norm": 0.12615702371946497,
      "learning_rate": 0.0001980281018175502,
      "loss": 1.0602,
      "step": 1269
    },
    {
      "epoch": 0.09154472716788005,
      "grad_norm": 0.3886502637866539,
      "learning_rate": 0.0001980234855500151,
      "loss": 1.0665,
      "step": 1270
    },
    {
      "epoch": 0.09161680963021697,
      "grad_norm": 0.1825155811090944,
      "learning_rate": 0.00019801886393932313,
      "loss": 1.0873,
      "step": 1271
    },
    {
      "epoch": 0.09168889209255388,
      "grad_norm": 0.13248462498363095,
      "learning_rate": 0.00019801423698572617,
      "loss": 1.2556,
      "step": 1272
    },
    {
      "epoch": 0.0917609745548908,
      "grad_norm": 0.10165441712755995,
      "learning_rate": 0.00019800960468947638,
      "loss": 1.1145,
      "step": 1273
    },
    {
      "epoch": 0.09183305701722771,
      "grad_norm": 0.12895657477744918,
      "learning_rate": 0.00019800496705082638,
      "loss": 1.1462,
      "step": 1274
    },
    {
      "epoch": 0.09190513947956462,
      "grad_norm": 0.1399261138349653,
      "learning_rate": 0.00019800032407002886,
      "loss": 1.1214,
      "step": 1275
    },
    {
      "epoch": 0.09197722194190154,
      "grad_norm": 0.25114180582202017,
      "learning_rate": 0.00019799567574733696,
      "loss": 1.1839,
      "step": 1276
    },
    {
      "epoch": 0.09204930440423845,
      "grad_norm": 0.17098645883440164,
      "learning_rate": 0.00019799102208300403,
      "loss": 1.1321,
      "step": 1277
    },
    {
      "epoch": 0.09212138686657537,
      "grad_norm": 0.13909842476069967,
      "learning_rate": 0.00019798636307728374,
      "loss": 1.0832,
      "step": 1278
    },
    {
      "epoch": 0.09219346932891227,
      "grad_norm": 0.15125687425877496,
      "learning_rate": 0.00019798169873043006,
      "loss": 1.0981,
      "step": 1279
    },
    {
      "epoch": 0.09226555179124919,
      "grad_norm": 0.13685392741826155,
      "learning_rate": 0.0001979770290426972,
      "loss": 1.1021,
      "step": 1280
    },
    {
      "epoch": 0.0923376342535861,
      "grad_norm": 0.259481254424911,
      "learning_rate": 0.00019797235401433977,
      "loss": 1.1534,
      "step": 1281
    },
    {
      "epoch": 0.09240971671592302,
      "grad_norm": 0.22827603765562363,
      "learning_rate": 0.00019796767364561256,
      "loss": 1.1182,
      "step": 1282
    },
    {
      "epoch": 0.09248179917825992,
      "grad_norm": 0.2102826281216397,
      "learning_rate": 0.00019796298793677068,
      "loss": 1.0603,
      "step": 1283
    },
    {
      "epoch": 0.09255388164059684,
      "grad_norm": 0.2627122822658727,
      "learning_rate": 0.00019795829688806955,
      "loss": 1.1916,
      "step": 1284
    },
    {
      "epoch": 0.09262596410293375,
      "grad_norm": 0.21967191446732373,
      "learning_rate": 0.00019795360049976486,
      "loss": 1.1811,
      "step": 1285
    },
    {
      "epoch": 0.09269804656527067,
      "grad_norm": 0.16084567628344074,
      "learning_rate": 0.00019794889877211268,
      "loss": 1.1538,
      "step": 1286
    },
    {
      "epoch": 0.09277012902760759,
      "grad_norm": 0.12293742210650004,
      "learning_rate": 0.00019794419170536916,
      "loss": 1.1878,
      "step": 1287
    },
    {
      "epoch": 0.0928422114899445,
      "grad_norm": 0.11969993266898443,
      "learning_rate": 0.00019793947929979105,
      "loss": 1.1254,
      "step": 1288
    },
    {
      "epoch": 0.09291429395228142,
      "grad_norm": 0.12282898008247736,
      "learning_rate": 0.00019793476155563507,
      "loss": 1.0755,
      "step": 1289
    },
    {
      "epoch": 0.09298637641461832,
      "grad_norm": 0.46711699379256255,
      "learning_rate": 0.00019793003847315844,
      "loss": 1.0973,
      "step": 1290
    },
    {
      "epoch": 0.09305845887695524,
      "grad_norm": 0.16081587423580443,
      "learning_rate": 0.00019792531005261862,
      "loss": 1.1513,
      "step": 1291
    },
    {
      "epoch": 0.09313054133929215,
      "grad_norm": 0.15254423570650616,
      "learning_rate": 0.00019792057629427335,
      "loss": 1.1466,
      "step": 1292
    },
    {
      "epoch": 0.09320262380162907,
      "grad_norm": 0.11352235153278749,
      "learning_rate": 0.0001979158371983806,
      "loss": 1.1163,
      "step": 1293
    },
    {
      "epoch": 0.09327470626396597,
      "grad_norm": 0.12152765202947245,
      "learning_rate": 0.00019791109276519878,
      "loss": 1.119,
      "step": 1294
    },
    {
      "epoch": 0.09334678872630289,
      "grad_norm": 0.13171841381501853,
      "learning_rate": 0.00019790634299498647,
      "loss": 1.0492,
      "step": 1295
    },
    {
      "epoch": 0.0934188711886398,
      "grad_norm": 0.17911669157519447,
      "learning_rate": 0.00019790158788800258,
      "loss": 1.0889,
      "step": 1296
    },
    {
      "epoch": 0.09349095365097672,
      "grad_norm": 0.15546784181177203,
      "learning_rate": 0.00019789682744450628,
      "loss": 1.0572,
      "step": 1297
    },
    {
      "epoch": 0.09356303611331362,
      "grad_norm": 0.11627396056234439,
      "learning_rate": 0.00019789206166475706,
      "loss": 1.1438,
      "step": 1298
    },
    {
      "epoch": 0.09363511857565054,
      "grad_norm": 0.2186692565811016,
      "learning_rate": 0.00019788729054901475,
      "loss": 1.1175,
      "step": 1299
    },
    {
      "epoch": 0.09370720103798746,
      "grad_norm": 0.3886337503697812,
      "learning_rate": 0.00019788251409753933,
      "loss": 1.125,
      "step": 1300
    },
    {
      "epoch": 0.09377928350032437,
      "grad_norm": 0.7320983768891507,
      "learning_rate": 0.00019787773231059124,
      "loss": 1.0811,
      "step": 1301
    },
    {
      "epoch": 0.09385136596266129,
      "grad_norm": 0.1344086010550082,
      "learning_rate": 0.00019787294518843108,
      "loss": 1.0988,
      "step": 1302
    },
    {
      "epoch": 0.0939234484249982,
      "grad_norm": 0.1296423588336384,
      "learning_rate": 0.00019786815273131986,
      "loss": 1.1307,
      "step": 1303
    },
    {
      "epoch": 0.09399553088733512,
      "grad_norm": 0.2211413688847426,
      "learning_rate": 0.00019786335493951872,
      "loss": 1.1498,
      "step": 1304
    },
    {
      "epoch": 0.09406761334967202,
      "grad_norm": 0.11238936178931344,
      "learning_rate": 0.00019785855181328924,
      "loss": 1.0899,
      "step": 1305
    },
    {
      "epoch": 0.09413969581200894,
      "grad_norm": 0.13579664426222246,
      "learning_rate": 0.00019785374335289321,
      "loss": 1.0914,
      "step": 1306
    },
    {
      "epoch": 0.09421177827434585,
      "grad_norm": 0.12353764005795288,
      "learning_rate": 0.00019784892955859276,
      "loss": 1.1411,
      "step": 1307
    },
    {
      "epoch": 0.09428386073668277,
      "grad_norm": 0.12350327850646121,
      "learning_rate": 0.00019784411043065022,
      "loss": 1.0133,
      "step": 1308
    },
    {
      "epoch": 0.09435594319901967,
      "grad_norm": 0.4181138771669546,
      "learning_rate": 0.00019783928596932838,
      "loss": 1.1784,
      "step": 1309
    },
    {
      "epoch": 0.0944280256613566,
      "grad_norm": 0.367925327759804,
      "learning_rate": 0.00019783445617489013,
      "loss": 1.0029,
      "step": 1310
    },
    {
      "epoch": 0.0945001081236935,
      "grad_norm": 0.18327281281048496,
      "learning_rate": 0.00019782962104759877,
      "loss": 1.1101,
      "step": 1311
    },
    {
      "epoch": 0.09457219058603042,
      "grad_norm": 0.7088860544916364,
      "learning_rate": 0.00019782478058771784,
      "loss": 1.123,
      "step": 1312
    },
    {
      "epoch": 0.09464427304836734,
      "grad_norm": 0.12774921643441053,
      "learning_rate": 0.00019781993479551125,
      "loss": 1.003,
      "step": 1313
    },
    {
      "epoch": 0.09471635551070424,
      "grad_norm": 0.22666820441273403,
      "learning_rate": 0.00019781508367124303,
      "loss": 1.1611,
      "step": 1314
    },
    {
      "epoch": 0.09478843797304116,
      "grad_norm": 0.35547353336164184,
      "learning_rate": 0.00019781022721517772,
      "loss": 1.0072,
      "step": 1315
    },
    {
      "epoch": 0.09486052043537807,
      "grad_norm": 0.11925373048053044,
      "learning_rate": 0.00019780536542758,
      "loss": 1.0126,
      "step": 1316
    },
    {
      "epoch": 0.09493260289771499,
      "grad_norm": 0.3561548728341217,
      "learning_rate": 0.00019780049830871483,
      "loss": 1.1457,
      "step": 1317
    },
    {
      "epoch": 0.0950046853600519,
      "grad_norm": 0.23253520996950014,
      "learning_rate": 0.00019779562585884756,
      "loss": 1.1381,
      "step": 1318
    },
    {
      "epoch": 0.09507676782238882,
      "grad_norm": 0.2164836305071356,
      "learning_rate": 0.00019779074807824382,
      "loss": 1.1818,
      "step": 1319
    },
    {
      "epoch": 0.09514885028472572,
      "grad_norm": 0.12236313151814411,
      "learning_rate": 0.0001977858649671694,
      "loss": 1.0695,
      "step": 1320
    },
    {
      "epoch": 0.09522093274706264,
      "grad_norm": 0.17673191646122174,
      "learning_rate": 0.00019778097652589053,
      "loss": 1.1478,
      "step": 1321
    },
    {
      "epoch": 0.09529301520939955,
      "grad_norm": 0.3324578519275603,
      "learning_rate": 0.00019777608275467365,
      "loss": 1.1068,
      "step": 1322
    },
    {
      "epoch": 0.09536509767173647,
      "grad_norm": 2.218936820764755,
      "learning_rate": 0.00019777118365378558,
      "loss": 1.134,
      "step": 1323
    },
    {
      "epoch": 0.09543718013407337,
      "grad_norm": 0.1199227121260367,
      "learning_rate": 0.0001977662792234933,
      "loss": 1.1848,
      "step": 1324
    },
    {
      "epoch": 0.0955092625964103,
      "grad_norm": 0.1219933097330172,
      "learning_rate": 0.00019776136946406413,
      "loss": 1.0272,
      "step": 1325
    },
    {
      "epoch": 0.09558134505874721,
      "grad_norm": 0.12422249303821299,
      "learning_rate": 0.00019775645437576574,
      "loss": 1.1175,
      "step": 1326
    },
    {
      "epoch": 0.09565342752108412,
      "grad_norm": 0.14469247520245773,
      "learning_rate": 0.00019775153395886603,
      "loss": 1.1772,
      "step": 1327
    },
    {
      "epoch": 0.09572550998342104,
      "grad_norm": 0.1913964848064828,
      "learning_rate": 0.0001977466082136332,
      "loss": 1.1186,
      "step": 1328
    },
    {
      "epoch": 0.09579759244575795,
      "grad_norm": 1.041494129585171,
      "learning_rate": 0.0001977416771403358,
      "loss": 1.1838,
      "step": 1329
    },
    {
      "epoch": 0.09586967490809487,
      "grad_norm": 0.1428296922534994,
      "learning_rate": 0.00019773674073924252,
      "loss": 1.1437,
      "step": 1330
    },
    {
      "epoch": 0.09594175737043177,
      "grad_norm": 0.3764701559074614,
      "learning_rate": 0.00019773179901062254,
      "loss": 1.1522,
      "step": 1331
    },
    {
      "epoch": 0.09601383983276869,
      "grad_norm": 0.11379289019803566,
      "learning_rate": 0.00019772685195474515,
      "loss": 1.1114,
      "step": 1332
    },
    {
      "epoch": 0.0960859222951056,
      "grad_norm": 0.24983220373756074,
      "learning_rate": 0.00019772189957188005,
      "loss": 1.0831,
      "step": 1333
    },
    {
      "epoch": 0.09615800475744252,
      "grad_norm": 0.9823566953447944,
      "learning_rate": 0.00019771694186229715,
      "loss": 1.039,
      "step": 1334
    },
    {
      "epoch": 0.09623008721977942,
      "grad_norm": 0.14415809232774923,
      "learning_rate": 0.00019771197882626678,
      "loss": 1.0648,
      "step": 1335
    },
    {
      "epoch": 0.09630216968211634,
      "grad_norm": 0.1396106777748474,
      "learning_rate": 0.00019770701046405934,
      "loss": 1.1208,
      "step": 1336
    },
    {
      "epoch": 0.09637425214445325,
      "grad_norm": 0.11866765386164829,
      "learning_rate": 0.00019770203677594577,
      "loss": 1.1324,
      "step": 1337
    },
    {
      "epoch": 0.09644633460679017,
      "grad_norm": 0.22332756075937715,
      "learning_rate": 0.00019769705776219709,
      "loss": 1.1698,
      "step": 1338
    },
    {
      "epoch": 0.09651841706912709,
      "grad_norm": 0.12748900602087462,
      "learning_rate": 0.00019769207342308477,
      "loss": 1.0958,
      "step": 1339
    },
    {
      "epoch": 0.096590499531464,
      "grad_norm": 0.1553150435743829,
      "learning_rate": 0.00019768708375888047,
      "loss": 1.1244,
      "step": 1340
    },
    {
      "epoch": 0.09666258199380091,
      "grad_norm": 0.14941757054675994,
      "learning_rate": 0.00019768208876985615,
      "loss": 1.0272,
      "step": 1341
    },
    {
      "epoch": 0.09673466445613782,
      "grad_norm": 0.13265154401833182,
      "learning_rate": 0.0001976770884562841,
      "loss": 1.0498,
      "step": 1342
    },
    {
      "epoch": 0.09680674691847474,
      "grad_norm": 0.3005390218993865,
      "learning_rate": 0.0001976720828184369,
      "loss": 1.1588,
      "step": 1343
    },
    {
      "epoch": 0.09687882938081165,
      "grad_norm": 0.2168596516023388,
      "learning_rate": 0.00019766707185658736,
      "loss": 1.1032,
      "step": 1344
    },
    {
      "epoch": 0.09695091184314857,
      "grad_norm": 0.14086407096490278,
      "learning_rate": 0.00019766205557100868,
      "loss": 1.1313,
      "step": 1345
    },
    {
      "epoch": 0.09702299430548547,
      "grad_norm": 0.13981205417453166,
      "learning_rate": 0.00019765703396197425,
      "loss": 1.114,
      "step": 1346
    },
    {
      "epoch": 0.09709507676782239,
      "grad_norm": 0.3511420058720167,
      "learning_rate": 0.00019765200702975778,
      "loss": 1.1736,
      "step": 1347
    },
    {
      "epoch": 0.0971671592301593,
      "grad_norm": 0.2254804285143275,
      "learning_rate": 0.00019764697477463338,
      "loss": 1.161,
      "step": 1348
    },
    {
      "epoch": 0.09723924169249622,
      "grad_norm": 0.1884680333198041,
      "learning_rate": 0.00019764193719687523,
      "loss": 1.0811,
      "step": 1349
    },
    {
      "epoch": 0.09731132415483312,
      "grad_norm": 0.13882633606677092,
      "learning_rate": 0.00019763689429675796,
      "loss": 0.9828,
      "step": 1350
    },
    {
      "epoch": 0.09738340661717004,
      "grad_norm": 0.3919715597576052,
      "learning_rate": 0.0001976318460745565,
      "loss": 1.0818,
      "step": 1351
    },
    {
      "epoch": 0.09745548907950696,
      "grad_norm": 0.15044450081868715,
      "learning_rate": 0.00019762679253054595,
      "loss": 1.1794,
      "step": 1352
    },
    {
      "epoch": 0.09752757154184387,
      "grad_norm": 0.11445387408025795,
      "learning_rate": 0.00019762173366500184,
      "loss": 1.1315,
      "step": 1353
    },
    {
      "epoch": 0.09759965400418079,
      "grad_norm": 0.1300554853373003,
      "learning_rate": 0.0001976166694781999,
      "loss": 1.0809,
      "step": 1354
    },
    {
      "epoch": 0.0976717364665177,
      "grad_norm": 0.9992323949183618,
      "learning_rate": 0.00019761159997041617,
      "loss": 1.1,
      "step": 1355
    },
    {
      "epoch": 0.09774381892885461,
      "grad_norm": 0.18955723615799036,
      "learning_rate": 0.00019760652514192698,
      "loss": 1.0673,
      "step": 1356
    },
    {
      "epoch": 0.09781590139119152,
      "grad_norm": 0.18129719519018658,
      "learning_rate": 0.00019760144499300893,
      "loss": 1.0952,
      "step": 1357
    },
    {
      "epoch": 0.09788798385352844,
      "grad_norm": 2.3500928459305372,
      "learning_rate": 0.000197596359523939,
      "loss": 1.1092,
      "step": 1358
    },
    {
      "epoch": 0.09796006631586535,
      "grad_norm": 0.21113527532325324,
      "learning_rate": 0.00019759126873499433,
      "loss": 1.1655,
      "step": 1359
    },
    {
      "epoch": 0.09803214877820227,
      "grad_norm": 0.1582635813715238,
      "learning_rate": 0.00019758617262645241,
      "loss": 1.176,
      "step": 1360
    },
    {
      "epoch": 0.09810423124053917,
      "grad_norm": 0.2548328403992578,
      "learning_rate": 0.0001975810711985911,
      "loss": 1.1115,
      "step": 1361
    },
    {
      "epoch": 0.09817631370287609,
      "grad_norm": 0.11389113115573114,
      "learning_rate": 0.00019757596445168843,
      "loss": 1.1044,
      "step": 1362
    },
    {
      "epoch": 0.098248396165213,
      "grad_norm": 0.15222389559321248,
      "learning_rate": 0.00019757085238602273,
      "loss": 1.0636,
      "step": 1363
    },
    {
      "epoch": 0.09832047862754992,
      "grad_norm": 0.32325541787349277,
      "learning_rate": 0.0001975657350018727,
      "loss": 1.1733,
      "step": 1364
    },
    {
      "epoch": 0.09839256108988684,
      "grad_norm": 0.11591823615942767,
      "learning_rate": 0.00019756061229951724,
      "loss": 1.1371,
      "step": 1365
    },
    {
      "epoch": 0.09846464355222374,
      "grad_norm": 0.18267046393143932,
      "learning_rate": 0.00019755548427923564,
      "loss": 1.1389,
      "step": 1366
    },
    {
      "epoch": 0.09853672601456066,
      "grad_norm": 0.10745289612458012,
      "learning_rate": 0.00019755035094130735,
      "loss": 1.0827,
      "step": 1367
    },
    {
      "epoch": 0.09860880847689757,
      "grad_norm": 0.32656238265728676,
      "learning_rate": 0.00019754521228601223,
      "loss": 1.2632,
      "step": 1368
    },
    {
      "epoch": 0.09868089093923449,
      "grad_norm": 0.153866934630582,
      "learning_rate": 0.00019754006831363042,
      "loss": 1.0884,
      "step": 1369
    },
    {
      "epoch": 0.0987529734015714,
      "grad_norm": 0.2576148610875898,
      "learning_rate": 0.0001975349190244422,
      "loss": 1.1666,
      "step": 1370
    },
    {
      "epoch": 0.09882505586390831,
      "grad_norm": 0.14368543175833415,
      "learning_rate": 0.0001975297644187284,
      "loss": 1.1953,
      "step": 1371
    },
    {
      "epoch": 0.09889713832624522,
      "grad_norm": 0.11838802743653619,
      "learning_rate": 0.00019752460449676983,
      "loss": 1.0749,
      "step": 1372
    },
    {
      "epoch": 0.09896922078858214,
      "grad_norm": 3.038140887852913,
      "learning_rate": 0.0001975194392588479,
      "loss": 1.1287,
      "step": 1373
    },
    {
      "epoch": 0.09904130325091905,
      "grad_norm": 0.13171550870717144,
      "learning_rate": 0.00019751426870524407,
      "loss": 1.0782,
      "step": 1374
    },
    {
      "epoch": 0.09911338571325597,
      "grad_norm": 0.4410314000848625,
      "learning_rate": 0.0001975090928362402,
      "loss": 1.1516,
      "step": 1375
    },
    {
      "epoch": 0.09918546817559287,
      "grad_norm": 0.4713067079275814,
      "learning_rate": 0.00019750391165211846,
      "loss": 1.0492,
      "step": 1376
    },
    {
      "epoch": 0.09925755063792979,
      "grad_norm": 0.3174845168141229,
      "learning_rate": 0.0001974987251531612,
      "loss": 1.0603,
      "step": 1377
    },
    {
      "epoch": 0.0993296331002667,
      "grad_norm": 0.14672133524490744,
      "learning_rate": 0.00019749353333965123,
      "loss": 1.1801,
      "step": 1378
    },
    {
      "epoch": 0.09940171556260362,
      "grad_norm": 0.3733310957690409,
      "learning_rate": 0.00019748833621187144,
      "loss": 1.0693,
      "step": 1379
    },
    {
      "epoch": 0.09947379802494054,
      "grad_norm": 1.2757748131354576,
      "learning_rate": 0.00019748313377010516,
      "loss": 1.0795,
      "step": 1380
    },
    {
      "epoch": 0.09954588048727744,
      "grad_norm": 0.14326367386222807,
      "learning_rate": 0.000197477926014636,
      "loss": 1.0544,
      "step": 1381
    },
    {
      "epoch": 0.09961796294961436,
      "grad_norm": 0.18807961505274048,
      "learning_rate": 0.00019747271294574782,
      "loss": 1.1562,
      "step": 1382
    },
    {
      "epoch": 0.09969004541195127,
      "grad_norm": 0.1828294360851243,
      "learning_rate": 0.00019746749456372474,
      "loss": 1.0205,
      "step": 1383
    },
    {
      "epoch": 0.09976212787428819,
      "grad_norm": 0.2788127513835101,
      "learning_rate": 0.0001974622708688513,
      "loss": 1.2139,
      "step": 1384
    },
    {
      "epoch": 0.0998342103366251,
      "grad_norm": 0.15678890855524302,
      "learning_rate": 0.00019745704186141213,
      "loss": 1.1251,
      "step": 1385
    },
    {
      "epoch": 0.09990629279896202,
      "grad_norm": 0.16092603743608577,
      "learning_rate": 0.00019745180754169228,
      "loss": 1.0729,
      "step": 1386
    },
    {
      "epoch": 0.09997837526129892,
      "grad_norm": 1.272712168530033,
      "learning_rate": 0.00019744656790997712,
      "loss": 1.1348,
      "step": 1387
    },
    {
      "epoch": 0.10005045772363584,
      "grad_norm": 0.23087191991958508,
      "learning_rate": 0.00019744132296655223,
      "loss": 1.1096,
      "step": 1388
    },
    {
      "epoch": 0.10012254018597275,
      "grad_norm": 0.2353503152212039,
      "learning_rate": 0.0001974360727117035,
      "loss": 1.2152,
      "step": 1389
    },
    {
      "epoch": 0.10019462264830967,
      "grad_norm": 0.31800053029962133,
      "learning_rate": 0.00019743081714571715,
      "loss": 1.1754,
      "step": 1390
    },
    {
      "epoch": 0.10026670511064657,
      "grad_norm": 0.24211823289987647,
      "learning_rate": 0.00019742555626887955,
      "loss": 1.2023,
      "step": 1391
    },
    {
      "epoch": 0.10033878757298349,
      "grad_norm": 0.13653392910154014,
      "learning_rate": 0.0001974202900814776,
      "loss": 1.2544,
      "step": 1392
    },
    {
      "epoch": 0.10041087003532041,
      "grad_norm": 0.22908423328297467,
      "learning_rate": 0.00019741501858379828,
      "loss": 1.1394,
      "step": 1393
    },
    {
      "epoch": 0.10048295249765732,
      "grad_norm": 0.5032119668766869,
      "learning_rate": 0.00019740974177612892,
      "loss": 1.1736,
      "step": 1394
    },
    {
      "epoch": 0.10055503495999424,
      "grad_norm": 0.1498920472089971,
      "learning_rate": 0.0001974044596587572,
      "loss": 1.1102,
      "step": 1395
    },
    {
      "epoch": 0.10062711742233114,
      "grad_norm": 0.378525908210381,
      "learning_rate": 0.00019739917223197105,
      "loss": 1.2416,
      "step": 1396
    },
    {
      "epoch": 0.10069919988466806,
      "grad_norm": 0.12908551179529154,
      "learning_rate": 0.00019739387949605864,
      "loss": 1.0388,
      "step": 1397
    },
    {
      "epoch": 0.10077128234700497,
      "grad_norm": 0.10847591330818378,
      "learning_rate": 0.00019738858145130846,
      "loss": 1.1149,
      "step": 1398
    },
    {
      "epoch": 0.10084336480934189,
      "grad_norm": 0.3311419006255597,
      "learning_rate": 0.00019738327809800933,
      "loss": 1.2179,
      "step": 1399
    },
    {
      "epoch": 0.1009154472716788,
      "grad_norm": 0.4696491130154828,
      "learning_rate": 0.00019737796943645038,
      "loss": 1.1116,
      "step": 1400
    },
    {
      "epoch": 0.10098752973401572,
      "grad_norm": 0.6359898521655237,
      "learning_rate": 0.0001973726554669209,
      "loss": 1.1484,
      "step": 1401
    },
    {
      "epoch": 0.10105961219635262,
      "grad_norm": 0.452122943880335,
      "learning_rate": 0.00019736733618971056,
      "loss": 1.1872,
      "step": 1402
    },
    {
      "epoch": 0.10113169465868954,
      "grad_norm": 0.16932813479387004,
      "learning_rate": 0.00019736201160510931,
      "loss": 1.1348,
      "step": 1403
    },
    {
      "epoch": 0.10120377712102645,
      "grad_norm": 0.7573247883801825,
      "learning_rate": 0.00019735668171340742,
      "loss": 1.1323,
      "step": 1404
    },
    {
      "epoch": 0.10127585958336337,
      "grad_norm": 0.1879255464350054,
      "learning_rate": 0.0001973513465148954,
      "loss": 1.1265,
      "step": 1405
    },
    {
      "epoch": 0.10134794204570029,
      "grad_norm": 0.3783713299354012,
      "learning_rate": 0.0001973460060098641,
      "loss": 1.1266,
      "step": 1406
    },
    {
      "epoch": 0.1014200245080372,
      "grad_norm": 0.15472704192726625,
      "learning_rate": 0.00019734066019860453,
      "loss": 1.0938,
      "step": 1407
    },
    {
      "epoch": 0.10149210697037411,
      "grad_norm": 0.3111021466812334,
      "learning_rate": 0.0001973353090814082,
      "loss": 1.1202,
      "step": 1408
    },
    {
      "epoch": 0.10156418943271102,
      "grad_norm": 0.09505693350650167,
      "learning_rate": 0.00019732995265856672,
      "loss": 1.0808,
      "step": 1409
    },
    {
      "epoch": 0.10163627189504794,
      "grad_norm": 0.12219470798217572,
      "learning_rate": 0.00019732459093037204,
      "loss": 1.0987,
      "step": 1410
    },
    {
      "epoch": 0.10170835435738484,
      "grad_norm": 0.5594766923368142,
      "learning_rate": 0.00019731922389711653,
      "loss": 1.1118,
      "step": 1411
    },
    {
      "epoch": 0.10178043681972176,
      "grad_norm": 0.3011442768838889,
      "learning_rate": 0.00019731385155909262,
      "loss": 1.1121,
      "step": 1412
    },
    {
      "epoch": 0.10185251928205867,
      "grad_norm": 1.926181754442876,
      "learning_rate": 0.0001973084739165932,
      "loss": 1.1635,
      "step": 1413
    },
    {
      "epoch": 0.10192460174439559,
      "grad_norm": 0.15165831560001816,
      "learning_rate": 0.00019730309096991147,
      "loss": 1.1151,
      "step": 1414
    },
    {
      "epoch": 0.1019966842067325,
      "grad_norm": 0.15448254327982086,
      "learning_rate": 0.00019729770271934073,
      "loss": 1.1609,
      "step": 1415
    },
    {
      "epoch": 0.10206876666906942,
      "grad_norm": 0.2825862234633132,
      "learning_rate": 0.00019729230916517478,
      "loss": 1.1677,
      "step": 1416
    },
    {
      "epoch": 0.10214084913140632,
      "grad_norm": 0.7552349670416704,
      "learning_rate": 0.00019728691030770757,
      "loss": 1.0876,
      "step": 1417
    },
    {
      "epoch": 0.10221293159374324,
      "grad_norm": 0.12602197414965544,
      "learning_rate": 0.00019728150614723336,
      "loss": 1.0664,
      "step": 1418
    },
    {
      "epoch": 0.10228501405608016,
      "grad_norm": 1.389583370766442,
      "learning_rate": 0.00019727609668404678,
      "loss": 1.1614,
      "step": 1419
    },
    {
      "epoch": 0.10235709651841707,
      "grad_norm": 0.26458886734864606,
      "learning_rate": 0.00019727068191844268,
      "loss": 1.1321,
      "step": 1420
    },
    {
      "epoch": 0.10242917898075399,
      "grad_norm": 1.5221018841799672,
      "learning_rate": 0.0001972652618507162,
      "loss": 1.1061,
      "step": 1421
    },
    {
      "epoch": 0.1025012614430909,
      "grad_norm": 0.10949690734586336,
      "learning_rate": 0.00019725983648116277,
      "loss": 1.1688,
      "step": 1422
    },
    {
      "epoch": 0.10257334390542781,
      "grad_norm": 0.8910979975827511,
      "learning_rate": 0.00019725440581007817,
      "loss": 1.1735,
      "step": 1423
    },
    {
      "epoch": 0.10264542636776472,
      "grad_norm": 0.2582456410296659,
      "learning_rate": 0.00019724896983775838,
      "loss": 1.0938,
      "step": 1424
    },
    {
      "epoch": 0.10271750883010164,
      "grad_norm": 0.9260497353617256,
      "learning_rate": 0.00019724352856449975,
      "loss": 1.1328,
      "step": 1425
    },
    {
      "epoch": 0.10278959129243855,
      "grad_norm": 0.463695937834462,
      "learning_rate": 0.00019723808199059882,
      "loss": 1.0874,
      "step": 1426
    },
    {
      "epoch": 0.10286167375477547,
      "grad_norm": 0.1265610086320453,
      "learning_rate": 0.0001972326301163525,
      "loss": 1.0783,
      "step": 1427
    },
    {
      "epoch": 0.10293375621711237,
      "grad_norm": 0.21553241741788792,
      "learning_rate": 0.00019722717294205796,
      "loss": 1.1267,
      "step": 1428
    },
    {
      "epoch": 0.10300583867944929,
      "grad_norm": 0.13886019107935071,
      "learning_rate": 0.0001972217104680127,
      "loss": 1.0506,
      "step": 1429
    },
    {
      "epoch": 0.1030779211417862,
      "grad_norm": 0.1462050884331332,
      "learning_rate": 0.00019721624269451445,
      "loss": 1.207,
      "step": 1430
    },
    {
      "epoch": 0.10315000360412312,
      "grad_norm": 0.18944576453852352,
      "learning_rate": 0.00019721076962186124,
      "loss": 1.1165,
      "step": 1431
    },
    {
      "epoch": 0.10322208606646004,
      "grad_norm": 0.15544551651006328,
      "learning_rate": 0.0001972052912503514,
      "loss": 1.0597,
      "step": 1432
    },
    {
      "epoch": 0.10329416852879694,
      "grad_norm": 0.17214065735340378,
      "learning_rate": 0.00019719980758028358,
      "loss": 1.0326,
      "step": 1433
    },
    {
      "epoch": 0.10336625099113386,
      "grad_norm": 0.9117877561933725,
      "learning_rate": 0.00019719431861195668,
      "loss": 1.0972,
      "step": 1434
    },
    {
      "epoch": 0.10343833345347077,
      "grad_norm": 0.4369088206546943,
      "learning_rate": 0.00019718882434566987,
      "loss": 1.1825,
      "step": 1435
    },
    {
      "epoch": 0.10351041591580769,
      "grad_norm": 0.13704713450204595,
      "learning_rate": 0.00019718332478172267,
      "loss": 1.1278,
      "step": 1436
    },
    {
      "epoch": 0.1035824983781446,
      "grad_norm": 0.13069957731670137,
      "learning_rate": 0.00019717781992041483,
      "loss": 1.1109,
      "step": 1437
    },
    {
      "epoch": 0.10365458084048151,
      "grad_norm": 0.17880803036537998,
      "learning_rate": 0.00019717230976204646,
      "loss": 1.0718,
      "step": 1438
    },
    {
      "epoch": 0.10372666330281842,
      "grad_norm": 0.1889739363160115,
      "learning_rate": 0.00019716679430691785,
      "loss": 1.1005,
      "step": 1439
    },
    {
      "epoch": 0.10379874576515534,
      "grad_norm": 0.13157329925694505,
      "learning_rate": 0.0001971612735553297,
      "loss": 1.1766,
      "step": 1440
    },
    {
      "epoch": 0.10387082822749225,
      "grad_norm": 0.2500225377045655,
      "learning_rate": 0.00019715574750758287,
      "loss": 1.0489,
      "step": 1441
    },
    {
      "epoch": 0.10394291068982917,
      "grad_norm": 0.1758538614531615,
      "learning_rate": 0.00019715021616397864,
      "loss": 1.0843,
      "step": 1442
    },
    {
      "epoch": 0.10401499315216607,
      "grad_norm": 0.14832622719544827,
      "learning_rate": 0.0001971446795248185,
      "loss": 1.1192,
      "step": 1443
    },
    {
      "epoch": 0.10408707561450299,
      "grad_norm": 0.12068137091045088,
      "learning_rate": 0.00019713913759040422,
      "loss": 1.1411,
      "step": 1444
    },
    {
      "epoch": 0.10415915807683991,
      "grad_norm": 0.18294266119190244,
      "learning_rate": 0.00019713359036103797,
      "loss": 1.0585,
      "step": 1445
    },
    {
      "epoch": 0.10423124053917682,
      "grad_norm": 0.11214323985234742,
      "learning_rate": 0.00019712803783702204,
      "loss": 1.0857,
      "step": 1446
    },
    {
      "epoch": 0.10430332300151374,
      "grad_norm": 0.12500533844235528,
      "learning_rate": 0.00019712248001865909,
      "loss": 1.1916,
      "step": 1447
    },
    {
      "epoch": 0.10437540546385064,
      "grad_norm": 0.14764960724984164,
      "learning_rate": 0.00019711691690625212,
      "loss": 1.1663,
      "step": 1448
    },
    {
      "epoch": 0.10444748792618756,
      "grad_norm": 0.4696573365978652,
      "learning_rate": 0.00019711134850010432,
      "loss": 1.1764,
      "step": 1449
    },
    {
      "epoch": 0.10451957038852447,
      "grad_norm": 0.11835799860204287,
      "learning_rate": 0.00019710577480051925,
      "loss": 1.1034,
      "step": 1450
    },
    {
      "epoch": 0.10459165285086139,
      "grad_norm": 0.12452402385371328,
      "learning_rate": 0.00019710019580780074,
      "loss": 1.1036,
      "step": 1451
    },
    {
      "epoch": 0.1046637353131983,
      "grad_norm": 0.3115134546064382,
      "learning_rate": 0.00019709461152225285,
      "loss": 1.1228,
      "step": 1452
    },
    {
      "epoch": 0.10473581777553521,
      "grad_norm": 0.12382122512679762,
      "learning_rate": 0.00019708902194418002,
      "loss": 1.1527,
      "step": 1453
    },
    {
      "epoch": 0.10480790023787212,
      "grad_norm": 0.18716741309676355,
      "learning_rate": 0.0001970834270738869,
      "loss": 1.1928,
      "step": 1454
    },
    {
      "epoch": 0.10487998270020904,
      "grad_norm": 0.11286622649415641,
      "learning_rate": 0.0001970778269116785,
      "loss": 1.116,
      "step": 1455
    },
    {
      "epoch": 0.10495206516254595,
      "grad_norm": 0.8093696972021525,
      "learning_rate": 0.00019707222145786,
      "loss": 1.2142,
      "step": 1456
    },
    {
      "epoch": 0.10502414762488287,
      "grad_norm": 0.12804121672069765,
      "learning_rate": 0.00019706661071273703,
      "loss": 1.139,
      "step": 1457
    },
    {
      "epoch": 0.10509623008721979,
      "grad_norm": 0.5413754209437713,
      "learning_rate": 0.00019706099467661542,
      "loss": 1.1692,
      "step": 1458
    },
    {
      "epoch": 0.10516831254955669,
      "grad_norm": 0.2850600735271855,
      "learning_rate": 0.0001970553733498012,
      "loss": 1.1385,
      "step": 1459
    },
    {
      "epoch": 0.10524039501189361,
      "grad_norm": 0.12447870646360135,
      "learning_rate": 0.00019704974673260091,
      "loss": 1.2016,
      "step": 1460
    },
    {
      "epoch": 0.10531247747423052,
      "grad_norm": 0.1926662599258428,
      "learning_rate": 0.00019704411482532116,
      "loss": 1.1367,
      "step": 1461
    },
    {
      "epoch": 0.10538455993656744,
      "grad_norm": 0.15162600235459023,
      "learning_rate": 0.00019703847762826896,
      "loss": 1.1027,
      "step": 1462
    },
    {
      "epoch": 0.10545664239890434,
      "grad_norm": 0.24042344695497422,
      "learning_rate": 0.00019703283514175163,
      "loss": 1.1563,
      "step": 1463
    },
    {
      "epoch": 0.10552872486124126,
      "grad_norm": 0.37734068749366634,
      "learning_rate": 0.00019702718736607668,
      "loss": 1.1106,
      "step": 1464
    },
    {
      "epoch": 0.10560080732357817,
      "grad_norm": 1.6055029821659668,
      "learning_rate": 0.000197021534301552,
      "loss": 1.1928,
      "step": 1465
    },
    {
      "epoch": 0.10567288978591509,
      "grad_norm": 0.12988937098732928,
      "learning_rate": 0.0001970158759484857,
      "loss": 1.0514,
      "step": 1466
    },
    {
      "epoch": 0.105744972248252,
      "grad_norm": 0.3626483790916902,
      "learning_rate": 0.00019701021230718628,
      "loss": 1.1453,
      "step": 1467
    },
    {
      "epoch": 0.10581705471058891,
      "grad_norm": 0.13014919109812717,
      "learning_rate": 0.00019700454337796236,
      "loss": 1.1349,
      "step": 1468
    },
    {
      "epoch": 0.10588913717292582,
      "grad_norm": 0.13819589222139422,
      "learning_rate": 0.00019699886916112305,
      "loss": 1.0867,
      "step": 1469
    },
    {
      "epoch": 0.10596121963526274,
      "grad_norm": 0.4762267457577917,
      "learning_rate": 0.00019699318965697756,
      "loss": 1.1972,
      "step": 1470
    },
    {
      "epoch": 0.10603330209759966,
      "grad_norm": 0.115301919085997,
      "learning_rate": 0.0001969875048658355,
      "loss": 1.1133,
      "step": 1471
    },
    {
      "epoch": 0.10610538455993657,
      "grad_norm": 0.3095007810504012,
      "learning_rate": 0.00019698181478800675,
      "loss": 1.1455,
      "step": 1472
    },
    {
      "epoch": 0.10617746702227349,
      "grad_norm": 0.13301341271488767,
      "learning_rate": 0.0001969761194238015,
      "loss": 1.1217,
      "step": 1473
    },
    {
      "epoch": 0.10624954948461039,
      "grad_norm": 0.31924583206097473,
      "learning_rate": 0.00019697041877353013,
      "loss": 1.1269,
      "step": 1474
    },
    {
      "epoch": 0.10632163194694731,
      "grad_norm": 0.22013358147768913,
      "learning_rate": 0.00019696471283750343,
      "loss": 1.1475,
      "step": 1475
    },
    {
      "epoch": 0.10639371440928422,
      "grad_norm": 0.5040008837413048,
      "learning_rate": 0.00019695900161603244,
      "loss": 1.0947,
      "step": 1476
    },
    {
      "epoch": 0.10646579687162114,
      "grad_norm": 0.19904097720574185,
      "learning_rate": 0.0001969532851094284,
      "loss": 1.1848,
      "step": 1477
    },
    {
      "epoch": 0.10653787933395804,
      "grad_norm": 0.44057547127751934,
      "learning_rate": 0.000196947563318003,
      "loss": 1.166,
      "step": 1478
    },
    {
      "epoch": 0.10660996179629496,
      "grad_norm": 0.3446771771591854,
      "learning_rate": 0.00019694183624206803,
      "loss": 1.2009,
      "step": 1479
    },
    {
      "epoch": 0.10668204425863187,
      "grad_norm": 0.11646235449938132,
      "learning_rate": 0.00019693610388193573,
      "loss": 1.1085,
      "step": 1480
    },
    {
      "epoch": 0.10675412672096879,
      "grad_norm": 0.1812451992983689,
      "learning_rate": 0.00019693036623791857,
      "loss": 1.0678,
      "step": 1481
    },
    {
      "epoch": 0.1068262091833057,
      "grad_norm": 0.1183914390912247,
      "learning_rate": 0.00019692462331032928,
      "loss": 1.0883,
      "step": 1482
    },
    {
      "epoch": 0.10689829164564262,
      "grad_norm": 0.13421849272706135,
      "learning_rate": 0.00019691887509948092,
      "loss": 1.0375,
      "step": 1483
    },
    {
      "epoch": 0.10697037410797954,
      "grad_norm": 1.3397758951644603,
      "learning_rate": 0.0001969131216056868,
      "loss": 1.1082,
      "step": 1484
    },
    {
      "epoch": 0.10704245657031644,
      "grad_norm": 0.13339914316360887,
      "learning_rate": 0.0001969073628292605,
      "loss": 1.1379,
      "step": 1485
    },
    {
      "epoch": 0.10711453903265336,
      "grad_norm": 0.18835962661411323,
      "learning_rate": 0.00019690159877051603,
      "loss": 1.0812,
      "step": 1486
    },
    {
      "epoch": 0.10718662149499027,
      "grad_norm": 0.11817598221687238,
      "learning_rate": 0.00019689582942976748,
      "loss": 1.138,
      "step": 1487
    },
    {
      "epoch": 0.10725870395732719,
      "grad_norm": 0.1948139230165441,
      "learning_rate": 0.00019689005480732938,
      "loss": 1.1625,
      "step": 1488
    },
    {
      "epoch": 0.10733078641966409,
      "grad_norm": 0.20360120945364663,
      "learning_rate": 0.00019688427490351652,
      "loss": 1.1223,
      "step": 1489
    },
    {
      "epoch": 0.10740286888200101,
      "grad_norm": 0.10535330142860537,
      "learning_rate": 0.00019687848971864389,
      "loss": 1.0564,
      "step": 1490
    },
    {
      "epoch": 0.10747495134433792,
      "grad_norm": 0.12215549805433704,
      "learning_rate": 0.00019687269925302685,
      "loss": 1.1675,
      "step": 1491
    },
    {
      "epoch": 0.10754703380667484,
      "grad_norm": 0.18501989628766125,
      "learning_rate": 0.00019686690350698109,
      "loss": 1.0971,
      "step": 1492
    },
    {
      "epoch": 0.10761911626901174,
      "grad_norm": 0.13850158800585888,
      "learning_rate": 0.0001968611024808225,
      "loss": 1.0446,
      "step": 1493
    },
    {
      "epoch": 0.10769119873134866,
      "grad_norm": 0.19816548963987216,
      "learning_rate": 0.00019685529617486728,
      "loss": 1.101,
      "step": 1494
    },
    {
      "epoch": 0.10776328119368557,
      "grad_norm": 0.15012015187017683,
      "learning_rate": 0.0001968494845894319,
      "loss": 1.1127,
      "step": 1495
    },
    {
      "epoch": 0.10783536365602249,
      "grad_norm": 0.1190836034757778,
      "learning_rate": 0.00019684366772483317,
      "loss": 1.0933,
      "step": 1496
    },
    {
      "epoch": 0.10790744611835941,
      "grad_norm": 0.10612738692860921,
      "learning_rate": 0.0001968378455813882,
      "loss": 1.0728,
      "step": 1497
    },
    {
      "epoch": 0.10797952858069632,
      "grad_norm": 0.11369827920342508,
      "learning_rate": 0.00019683201815941427,
      "loss": 1.107,
      "step": 1498
    },
    {
      "epoch": 0.10805161104303324,
      "grad_norm": 0.11515152077991984,
      "learning_rate": 0.00019682618545922907,
      "loss": 1.0661,
      "step": 1499
    },
    {
      "epoch": 0.10812369350537014,
      "grad_norm": 0.22632566200500434,
      "learning_rate": 0.00019682034748115055,
      "loss": 1.0998,
      "step": 1500
    },
    {
      "epoch": 0.10819577596770706,
      "grad_norm": 0.21994308715567676,
      "learning_rate": 0.0001968145042254969,
      "loss": 1.0911,
      "step": 1501
    },
    {
      "epoch": 0.10826785843004397,
      "grad_norm": 0.1443435954255445,
      "learning_rate": 0.00019680865569258663,
      "loss": 1.0997,
      "step": 1502
    },
    {
      "epoch": 0.10833994089238089,
      "grad_norm": 0.22655614121887313,
      "learning_rate": 0.00019680280188273859,
      "loss": 1.1013,
      "step": 1503
    },
    {
      "epoch": 0.10841202335471779,
      "grad_norm": 0.14866344861336883,
      "learning_rate": 0.00019679694279627177,
      "loss": 1.0698,
      "step": 1504
    },
    {
      "epoch": 0.10848410581705471,
      "grad_norm": 0.1426431421540061,
      "learning_rate": 0.00019679107843350563,
      "loss": 1.1428,
      "step": 1505
    },
    {
      "epoch": 0.10855618827939162,
      "grad_norm": 0.12456944928860099,
      "learning_rate": 0.0001967852087947598,
      "loss": 1.1585,
      "step": 1506
    },
    {
      "epoch": 0.10862827074172854,
      "grad_norm": 0.11746071251694791,
      "learning_rate": 0.0001967793338803542,
      "loss": 1.137,
      "step": 1507
    },
    {
      "epoch": 0.10870035320406544,
      "grad_norm": 0.25623562926038385,
      "learning_rate": 0.00019677345369060914,
      "loss": 1.1467,
      "step": 1508
    },
    {
      "epoch": 0.10877243566640236,
      "grad_norm": 0.2607281078320205,
      "learning_rate": 0.00019676756822584505,
      "loss": 1.0816,
      "step": 1509
    },
    {
      "epoch": 0.10884451812873928,
      "grad_norm": 0.11249860556618063,
      "learning_rate": 0.0001967616774863828,
      "loss": 1.0777,
      "step": 1510
    },
    {
      "epoch": 0.10891660059107619,
      "grad_norm": 0.11657309620857263,
      "learning_rate": 0.00019675578147254348,
      "loss": 1.054,
      "step": 1511
    },
    {
      "epoch": 0.10898868305341311,
      "grad_norm": 0.11170076289756767,
      "learning_rate": 0.00019674988018464847,
      "loss": 1.1081,
      "step": 1512
    },
    {
      "epoch": 0.10906076551575002,
      "grad_norm": 0.1397931951323087,
      "learning_rate": 0.00019674397362301943,
      "loss": 1.0906,
      "step": 1513
    },
    {
      "epoch": 0.10913284797808694,
      "grad_norm": 2.1484658363838953,
      "learning_rate": 0.00019673806178797834,
      "loss": 0.9995,
      "step": 1514
    },
    {
      "epoch": 0.10920493044042384,
      "grad_norm": 0.1264259481499402,
      "learning_rate": 0.00019673214467984742,
      "loss": 1.1688,
      "step": 1515
    },
    {
      "epoch": 0.10927701290276076,
      "grad_norm": 0.1496032225021935,
      "learning_rate": 0.00019672622229894923,
      "loss": 1.0862,
      "step": 1516
    },
    {
      "epoch": 0.10934909536509767,
      "grad_norm": 0.8186472803673097,
      "learning_rate": 0.0001967202946456066,
      "loss": 1.1831,
      "step": 1517
    },
    {
      "epoch": 0.10942117782743459,
      "grad_norm": 0.1223192052815275,
      "learning_rate": 0.00019671436172014262,
      "loss": 1.1039,
      "step": 1518
    },
    {
      "epoch": 0.1094932602897715,
      "grad_norm": 0.13561204547755137,
      "learning_rate": 0.0001967084235228807,
      "loss": 1.1113,
      "step": 1519
    },
    {
      "epoch": 0.10956534275210841,
      "grad_norm": 0.1637780395023837,
      "learning_rate": 0.0001967024800541445,
      "loss": 1.1118,
      "step": 1520
    },
    {
      "epoch": 0.10963742521444532,
      "grad_norm": 0.25907553544618234,
      "learning_rate": 0.00019669653131425803,
      "loss": 1.1064,
      "step": 1521
    },
    {
      "epoch": 0.10970950767678224,
      "grad_norm": 0.11563202186826835,
      "learning_rate": 0.00019669057730354553,
      "loss": 1.0683,
      "step": 1522
    },
    {
      "epoch": 0.10978159013911916,
      "grad_norm": 0.1300918712722564,
      "learning_rate": 0.00019668461802233153,
      "loss": 1.1148,
      "step": 1523
    },
    {
      "epoch": 0.10985367260145606,
      "grad_norm": 0.19320576941061693,
      "learning_rate": 0.00019667865347094092,
      "loss": 1.0966,
      "step": 1524
    },
    {
      "epoch": 0.10992575506379298,
      "grad_norm": 0.12867184710485277,
      "learning_rate": 0.00019667268364969876,
      "loss": 1.1661,
      "step": 1525
    },
    {
      "epoch": 0.10999783752612989,
      "grad_norm": 0.1953574149189956,
      "learning_rate": 0.00019666670855893049,
      "loss": 1.1611,
      "step": 1526
    },
    {
      "epoch": 0.11006991998846681,
      "grad_norm": 0.1896335162850818,
      "learning_rate": 0.00019666072819896177,
      "loss": 1.1443,
      "step": 1527
    },
    {
      "epoch": 0.11014200245080372,
      "grad_norm": 0.11706552313994414,
      "learning_rate": 0.00019665474257011864,
      "loss": 1.1245,
      "step": 1528
    },
    {
      "epoch": 0.11021408491314064,
      "grad_norm": 0.1206937153936934,
      "learning_rate": 0.00019664875167272735,
      "loss": 1.1546,
      "step": 1529
    },
    {
      "epoch": 0.11028616737547754,
      "grad_norm": 0.14062730615415858,
      "learning_rate": 0.0001966427555071144,
      "loss": 1.129,
      "step": 1530
    },
    {
      "epoch": 0.11035824983781446,
      "grad_norm": 0.15315428098806713,
      "learning_rate": 0.00019663675407360676,
      "loss": 1.0873,
      "step": 1531
    },
    {
      "epoch": 0.11043033230015137,
      "grad_norm": 0.21247638815105901,
      "learning_rate": 0.00019663074737253142,
      "loss": 1.0612,
      "step": 1532
    },
    {
      "epoch": 0.11050241476248829,
      "grad_norm": 0.19129638917437058,
      "learning_rate": 0.00019662473540421588,
      "loss": 1.0789,
      "step": 1533
    },
    {
      "epoch": 0.1105744972248252,
      "grad_norm": 0.11494822864810227,
      "learning_rate": 0.00019661871816898784,
      "loss": 1.1198,
      "step": 1534
    },
    {
      "epoch": 0.11064657968716211,
      "grad_norm": 0.12573879782449715,
      "learning_rate": 0.0001966126956671753,
      "loss": 1.1649,
      "step": 1535
    },
    {
      "epoch": 0.11071866214949903,
      "grad_norm": 0.14657461490691293,
      "learning_rate": 0.00019660666789910654,
      "loss": 1.1258,
      "step": 1536
    },
    {
      "epoch": 0.11079074461183594,
      "grad_norm": 0.11681342974248253,
      "learning_rate": 0.00019660063486511006,
      "loss": 1.132,
      "step": 1537
    },
    {
      "epoch": 0.11086282707417286,
      "grad_norm": 1.0856681840156235,
      "learning_rate": 0.0001965945965655148,
      "loss": 1.1339,
      "step": 1538
    },
    {
      "epoch": 0.11093490953650977,
      "grad_norm": 0.1584535628309368,
      "learning_rate": 0.00019658855300064984,
      "loss": 1.1218,
      "step": 1539
    },
    {
      "epoch": 0.11100699199884669,
      "grad_norm": 0.1357976969746976,
      "learning_rate": 0.00019658250417084466,
      "loss": 1.1092,
      "step": 1540
    },
    {
      "epoch": 0.11107907446118359,
      "grad_norm": 0.20697735978806192,
      "learning_rate": 0.00019657645007642894,
      "loss": 1.1045,
      "step": 1541
    },
    {
      "epoch": 0.11115115692352051,
      "grad_norm": 0.16376714425554528,
      "learning_rate": 0.0001965703907177327,
      "loss": 1.1275,
      "step": 1542
    },
    {
      "epoch": 0.11122323938585742,
      "grad_norm": 0.12080569488571312,
      "learning_rate": 0.0001965643260950862,
      "loss": 1.0174,
      "step": 1543
    },
    {
      "epoch": 0.11129532184819434,
      "grad_norm": 0.14930477826074598,
      "learning_rate": 0.00019655825620882005,
      "loss": 1.2063,
      "step": 1544
    },
    {
      "epoch": 0.11136740431053124,
      "grad_norm": 0.13740262630181882,
      "learning_rate": 0.0001965521810592651,
      "loss": 1.103,
      "step": 1545
    },
    {
      "epoch": 0.11143948677286816,
      "grad_norm": 0.1348021829945887,
      "learning_rate": 0.00019654610064675251,
      "loss": 1.1359,
      "step": 1546
    },
    {
      "epoch": 0.11151156923520507,
      "grad_norm": 0.12230830853318342,
      "learning_rate": 0.00019654001497161372,
      "loss": 1.1139,
      "step": 1547
    },
    {
      "epoch": 0.11158365169754199,
      "grad_norm": 0.12020675812309384,
      "learning_rate": 0.00019653392403418043,
      "loss": 1.1447,
      "step": 1548
    },
    {
      "epoch": 0.11165573415987891,
      "grad_norm": 0.13290745577287952,
      "learning_rate": 0.00019652782783478466,
      "loss": 1.0721,
      "step": 1549
    },
    {
      "epoch": 0.11172781662221581,
      "grad_norm": 0.10936975885453332,
      "learning_rate": 0.0001965217263737587,
      "loss": 1.0765,
      "step": 1550
    },
    {
      "epoch": 0.11179989908455273,
      "grad_norm": 0.5001823473817131,
      "learning_rate": 0.0001965156196514352,
      "loss": 1.0447,
      "step": 1551
    },
    {
      "epoch": 0.11187198154688964,
      "grad_norm": 0.1396464082666847,
      "learning_rate": 0.0001965095076681469,
      "loss": 1.0896,
      "step": 1552
    },
    {
      "epoch": 0.11194406400922656,
      "grad_norm": 0.35053400546066327,
      "learning_rate": 0.00019650339042422707,
      "loss": 1.0292,
      "step": 1553
    },
    {
      "epoch": 0.11201614647156347,
      "grad_norm": 0.10087654965608127,
      "learning_rate": 0.00019649726792000913,
      "loss": 1.1022,
      "step": 1554
    },
    {
      "epoch": 0.11208822893390039,
      "grad_norm": 0.21699390371964564,
      "learning_rate": 0.00019649114015582678,
      "loss": 1.1606,
      "step": 1555
    },
    {
      "epoch": 0.11216031139623729,
      "grad_norm": 0.114362181354937,
      "learning_rate": 0.00019648500713201404,
      "loss": 1.1073,
      "step": 1556
    },
    {
      "epoch": 0.11223239385857421,
      "grad_norm": 0.1414399501350991,
      "learning_rate": 0.00019647886884890527,
      "loss": 1.2126,
      "step": 1557
    },
    {
      "epoch": 0.11230447632091112,
      "grad_norm": 0.17669034811180653,
      "learning_rate": 0.00019647272530683502,
      "loss": 1.1153,
      "step": 1558
    },
    {
      "epoch": 0.11237655878324804,
      "grad_norm": 0.2258934007277283,
      "learning_rate": 0.00019646657650613817,
      "loss": 1.0844,
      "step": 1559
    },
    {
      "epoch": 0.11244864124558494,
      "grad_norm": 0.13835434562791213,
      "learning_rate": 0.00019646042244714985,
      "loss": 1.1209,
      "step": 1560
    },
    {
      "epoch": 0.11252072370792186,
      "grad_norm": 0.16451763376392434,
      "learning_rate": 0.00019645426313020557,
      "loss": 1.0675,
      "step": 1561
    },
    {
      "epoch": 0.11259280617025878,
      "grad_norm": 0.150585164023018,
      "learning_rate": 0.00019644809855564102,
      "loss": 1.0984,
      "step": 1562
    },
    {
      "epoch": 0.11266488863259569,
      "grad_norm": 0.13852604243874747,
      "learning_rate": 0.00019644192872379229,
      "loss": 1.1673,
      "step": 1563
    },
    {
      "epoch": 0.11273697109493261,
      "grad_norm": 0.12045483262604914,
      "learning_rate": 0.00019643575363499562,
      "loss": 1.0598,
      "step": 1564
    },
    {
      "epoch": 0.11280905355726951,
      "grad_norm": 0.12312911971156608,
      "learning_rate": 0.00019642957328958768,
      "loss": 1.168,
      "step": 1565
    },
    {
      "epoch": 0.11288113601960643,
      "grad_norm": 0.12220405262858133,
      "learning_rate": 0.00019642338768790526,
      "loss": 1.1147,
      "step": 1566
    },
    {
      "epoch": 0.11295321848194334,
      "grad_norm": 0.12232420542367498,
      "learning_rate": 0.0001964171968302856,
      "loss": 1.0227,
      "step": 1567
    },
    {
      "epoch": 0.11302530094428026,
      "grad_norm": 0.11008698862395512,
      "learning_rate": 0.00019641100071706618,
      "loss": 1.1093,
      "step": 1568
    },
    {
      "epoch": 0.11309738340661717,
      "grad_norm": 0.15626440851465,
      "learning_rate": 0.00019640479934858465,
      "loss": 1.0629,
      "step": 1569
    },
    {
      "epoch": 0.11316946586895409,
      "grad_norm": 0.1457730732052885,
      "learning_rate": 0.0001963985927251791,
      "loss": 1.0441,
      "step": 1570
    },
    {
      "epoch": 0.11324154833129099,
      "grad_norm": 0.1422477600057448,
      "learning_rate": 0.00019639238084718787,
      "loss": 1.0816,
      "step": 1571
    },
    {
      "epoch": 0.11331363079362791,
      "grad_norm": 0.13680702532091443,
      "learning_rate": 0.0001963861637149495,
      "loss": 1.0948,
      "step": 1572
    },
    {
      "epoch": 0.11338571325596482,
      "grad_norm": 0.11046969033873298,
      "learning_rate": 0.00019637994132880296,
      "loss": 1.076,
      "step": 1573
    },
    {
      "epoch": 0.11345779571830174,
      "grad_norm": 0.15593506278215868,
      "learning_rate": 0.00019637371368908736,
      "loss": 1.0594,
      "step": 1574
    },
    {
      "epoch": 0.11352987818063866,
      "grad_norm": 0.13214978547102196,
      "learning_rate": 0.00019636748079614216,
      "loss": 1.1372,
      "step": 1575
    },
    {
      "epoch": 0.11360196064297556,
      "grad_norm": 0.12090812235012895,
      "learning_rate": 0.00019636124265030714,
      "loss": 1.1147,
      "step": 1576
    },
    {
      "epoch": 0.11367404310531248,
      "grad_norm": 0.2997653595058523,
      "learning_rate": 0.0001963549992519223,
      "loss": 1.1998,
      "step": 1577
    },
    {
      "epoch": 0.11374612556764939,
      "grad_norm": 0.12101365657255378,
      "learning_rate": 0.000196348750601328,
      "loss": 1.21,
      "step": 1578
    },
    {
      "epoch": 0.11381820802998631,
      "grad_norm": 0.1196775111583199,
      "learning_rate": 0.00019634249669886483,
      "loss": 1.104,
      "step": 1579
    },
    {
      "epoch": 0.11389029049232322,
      "grad_norm": 0.1457741442832033,
      "learning_rate": 0.00019633623754487372,
      "loss": 1.0765,
      "step": 1580
    },
    {
      "epoch": 0.11396237295466013,
      "grad_norm": 0.13450426042455652,
      "learning_rate": 0.00019632997313969578,
      "loss": 1.1091,
      "step": 1581
    },
    {
      "epoch": 0.11403445541699704,
      "grad_norm": 0.1250229458030189,
      "learning_rate": 0.0001963237034836725,
      "loss": 1.1189,
      "step": 1582
    },
    {
      "epoch": 0.11410653787933396,
      "grad_norm": 0.10725675964866072,
      "learning_rate": 0.00019631742857714567,
      "loss": 1.156,
      "step": 1583
    },
    {
      "epoch": 0.11417862034167087,
      "grad_norm": 0.14964480858759047,
      "learning_rate": 0.0001963111484204573,
      "loss": 1.1354,
      "step": 1584
    },
    {
      "epoch": 0.11425070280400779,
      "grad_norm": 0.12132258024139007,
      "learning_rate": 0.0001963048630139497,
      "loss": 1.1652,
      "step": 1585
    },
    {
      "epoch": 0.11432278526634469,
      "grad_norm": 0.11101463661113668,
      "learning_rate": 0.00019629857235796552,
      "loss": 1.1443,
      "step": 1586
    },
    {
      "epoch": 0.11439486772868161,
      "grad_norm": 0.6676840507730507,
      "learning_rate": 0.0001962922764528476,
      "loss": 1.1421,
      "step": 1587
    },
    {
      "epoch": 0.11446695019101852,
      "grad_norm": 0.13504371378107993,
      "learning_rate": 0.0001962859752989392,
      "loss": 1.1515,
      "step": 1588
    },
    {
      "epoch": 0.11453903265335544,
      "grad_norm": 0.12440562766120837,
      "learning_rate": 0.00019627966889658373,
      "loss": 1.0846,
      "step": 1589
    },
    {
      "epoch": 0.11461111511569236,
      "grad_norm": 0.19315819868907375,
      "learning_rate": 0.00019627335724612494,
      "loss": 1.1548,
      "step": 1590
    },
    {
      "epoch": 0.11468319757802926,
      "grad_norm": 0.1972263362876483,
      "learning_rate": 0.00019626704034790692,
      "loss": 1.109,
      "step": 1591
    },
    {
      "epoch": 0.11475528004036618,
      "grad_norm": 0.2033675920928696,
      "learning_rate": 0.000196260718202274,
      "loss": 1.0573,
      "step": 1592
    },
    {
      "epoch": 0.11482736250270309,
      "grad_norm": 0.16063677639744528,
      "learning_rate": 0.0001962543908095707,
      "loss": 1.1483,
      "step": 1593
    },
    {
      "epoch": 0.11489944496504001,
      "grad_norm": 0.12282983551198863,
      "learning_rate": 0.000196248058170142,
      "loss": 1.0796,
      "step": 1594
    },
    {
      "epoch": 0.11497152742737692,
      "grad_norm": 0.2306615433195375,
      "learning_rate": 0.0001962417202843331,
      "loss": 1.1531,
      "step": 1595
    },
    {
      "epoch": 0.11504360988971384,
      "grad_norm": 0.10247283002615597,
      "learning_rate": 0.00019623537715248942,
      "loss": 1.1166,
      "step": 1596
    },
    {
      "epoch": 0.11511569235205074,
      "grad_norm": 0.1644605040319592,
      "learning_rate": 0.00019622902877495678,
      "loss": 1.1226,
      "step": 1597
    },
    {
      "epoch": 0.11518777481438766,
      "grad_norm": 0.11598401195721629,
      "learning_rate": 0.00019622267515208113,
      "loss": 1.1614,
      "step": 1598
    },
    {
      "epoch": 0.11525985727672457,
      "grad_norm": 0.12496737655785717,
      "learning_rate": 0.00019621631628420887,
      "loss": 1.1252,
      "step": 1599
    },
    {
      "epoch": 0.11533193973906149,
      "grad_norm": 1.1621235791017182,
      "learning_rate": 0.00019620995217168658,
      "loss": 1.0841,
      "step": 1600
    },
    {
      "epoch": 0.11540402220139839,
      "grad_norm": 0.10735421546964977,
      "learning_rate": 0.0001962035828148612,
      "loss": 1.1562,
      "step": 1601
    },
    {
      "epoch": 0.11547610466373531,
      "grad_norm": 0.5836233464942953,
      "learning_rate": 0.0001961972082140799,
      "loss": 1.0659,
      "step": 1602
    },
    {
      "epoch": 0.11554818712607223,
      "grad_norm": 0.09923012685293114,
      "learning_rate": 0.00019619082836969015,
      "loss": 1.1028,
      "step": 1603
    },
    {
      "epoch": 0.11562026958840914,
      "grad_norm": 0.10488692396945616,
      "learning_rate": 0.00019618444328203967,
      "loss": 1.0514,
      "step": 1604
    },
    {
      "epoch": 0.11569235205074606,
      "grad_norm": 0.2575619255192133,
      "learning_rate": 0.00019617805295147654,
      "loss": 1.0676,
      "step": 1605
    },
    {
      "epoch": 0.11576443451308296,
      "grad_norm": 0.14446887092005026,
      "learning_rate": 0.00019617165737834916,
      "loss": 1.1652,
      "step": 1606
    },
    {
      "epoch": 0.11583651697541988,
      "grad_norm": 0.2512279066775059,
      "learning_rate": 0.00019616525656300602,
      "loss": 1.1543,
      "step": 1607
    },
    {
      "epoch": 0.11590859943775679,
      "grad_norm": 1.4004059310924535,
      "learning_rate": 0.00019615885050579607,
      "loss": 1.1983,
      "step": 1608
    },
    {
      "epoch": 0.11598068190009371,
      "grad_norm": 1.09723731930983,
      "learning_rate": 0.00019615243920706853,
      "loss": 1.0682,
      "step": 1609
    },
    {
      "epoch": 0.11605276436243062,
      "grad_norm": 0.43680366921567787,
      "learning_rate": 0.00019614602266717284,
      "loss": 1.1869,
      "step": 1610
    },
    {
      "epoch": 0.11612484682476754,
      "grad_norm": 0.11705262141737222,
      "learning_rate": 0.00019613960088645877,
      "loss": 1.095,
      "step": 1611
    },
    {
      "epoch": 0.11619692928710444,
      "grad_norm": 0.12875496964825967,
      "learning_rate": 0.00019613317386527637,
      "loss": 1.0983,
      "step": 1612
    },
    {
      "epoch": 0.11626901174944136,
      "grad_norm": 0.20250077865597335,
      "learning_rate": 0.00019612674160397595,
      "loss": 1.181,
      "step": 1613
    },
    {
      "epoch": 0.11634109421177827,
      "grad_norm": 0.1426245306532939,
      "learning_rate": 0.00019612030410290812,
      "loss": 1.1141,
      "step": 1614
    },
    {
      "epoch": 0.11641317667411519,
      "grad_norm": 0.14481296393183288,
      "learning_rate": 0.00019611386136242386,
      "loss": 1.1437,
      "step": 1615
    },
    {
      "epoch": 0.11648525913645211,
      "grad_norm": 0.1440693537259232,
      "learning_rate": 0.00019610741338287423,
      "loss": 1.1418,
      "step": 1616
    },
    {
      "epoch": 0.11655734159878901,
      "grad_norm": 0.13579750380483993,
      "learning_rate": 0.00019610096016461085,
      "loss": 1.1082,
      "step": 1617
    },
    {
      "epoch": 0.11662942406112593,
      "grad_norm": 0.14868554112151713,
      "learning_rate": 0.00019609450170798533,
      "loss": 1.0752,
      "step": 1618
    },
    {
      "epoch": 0.11670150652346284,
      "grad_norm": 0.10348319317229522,
      "learning_rate": 0.0001960880380133498,
      "loss": 1.1015,
      "step": 1619
    },
    {
      "epoch": 0.11677358898579976,
      "grad_norm": 0.2586016285409722,
      "learning_rate": 0.00019608156908105655,
      "loss": 1.0944,
      "step": 1620
    },
    {
      "epoch": 0.11684567144813666,
      "grad_norm": 0.11868422823918458,
      "learning_rate": 0.00019607509491145823,
      "loss": 1.1578,
      "step": 1621
    },
    {
      "epoch": 0.11691775391047358,
      "grad_norm": 0.12589318510833167,
      "learning_rate": 0.00019606861550490774,
      "loss": 1.1,
      "step": 1622
    },
    {
      "epoch": 0.11698983637281049,
      "grad_norm": 0.1606488603143175,
      "learning_rate": 0.00019606213086175825,
      "loss": 1.1068,
      "step": 1623
    },
    {
      "epoch": 0.11706191883514741,
      "grad_norm": 0.5715187834012022,
      "learning_rate": 0.0001960556409823632,
      "loss": 1.0707,
      "step": 1624
    },
    {
      "epoch": 0.11713400129748432,
      "grad_norm": 0.3915594779099628,
      "learning_rate": 0.0001960491458670764,
      "loss": 1.0147,
      "step": 1625
    },
    {
      "epoch": 0.11720608375982124,
      "grad_norm": 0.11366897705861688,
      "learning_rate": 0.00019604264551625187,
      "loss": 1.1457,
      "step": 1626
    },
    {
      "epoch": 0.11727816622215814,
      "grad_norm": 0.15499991771652794,
      "learning_rate": 0.00019603613993024395,
      "loss": 1.101,
      "step": 1627
    },
    {
      "epoch": 0.11735024868449506,
      "grad_norm": 0.3434986322792138,
      "learning_rate": 0.00019602962910940722,
      "loss": 1.0723,
      "step": 1628
    },
    {
      "epoch": 0.11742233114683198,
      "grad_norm": 0.2887506569301316,
      "learning_rate": 0.0001960231130540966,
      "loss": 1.1405,
      "step": 1629
    },
    {
      "epoch": 0.11749441360916889,
      "grad_norm": 0.13829496230524582,
      "learning_rate": 0.00019601659176466727,
      "loss": 1.0524,
      "step": 1630
    },
    {
      "epoch": 0.11756649607150581,
      "grad_norm": 0.10670215885833739,
      "learning_rate": 0.0001960100652414747,
      "loss": 1.0643,
      "step": 1631
    },
    {
      "epoch": 0.11763857853384271,
      "grad_norm": 1.686037005000265,
      "learning_rate": 0.00019600353348487464,
      "loss": 1.1388,
      "step": 1632
    },
    {
      "epoch": 0.11771066099617963,
      "grad_norm": 0.13935258462439484,
      "learning_rate": 0.00019599699649522316,
      "loss": 1.1466,
      "step": 1633
    },
    {
      "epoch": 0.11778274345851654,
      "grad_norm": 0.1292749399011905,
      "learning_rate": 0.00019599045427287653,
      "loss": 1.1882,
      "step": 1634
    },
    {
      "epoch": 0.11785482592085346,
      "grad_norm": 1.3488134128682125,
      "learning_rate": 0.0001959839068181914,
      "loss": 1.0894,
      "step": 1635
    },
    {
      "epoch": 0.11792690838319037,
      "grad_norm": 0.1955474459592688,
      "learning_rate": 0.00019597735413152463,
      "loss": 1.0915,
      "step": 1636
    },
    {
      "epoch": 0.11799899084552729,
      "grad_norm": 0.13969149726990532,
      "learning_rate": 0.00019597079621323346,
      "loss": 1.1335,
      "step": 1637
    },
    {
      "epoch": 0.11807107330786419,
      "grad_norm": 0.11070407934853242,
      "learning_rate": 0.0001959642330636753,
      "loss": 1.0745,
      "step": 1638
    },
    {
      "epoch": 0.11814315577020111,
      "grad_norm": 0.3175625883434126,
      "learning_rate": 0.00019595766468320788,
      "loss": 1.0567,
      "step": 1639
    },
    {
      "epoch": 0.11821523823253802,
      "grad_norm": 0.1427911497236013,
      "learning_rate": 0.00019595109107218932,
      "loss": 1.1384,
      "step": 1640
    },
    {
      "epoch": 0.11828732069487494,
      "grad_norm": 1.0405672139846227,
      "learning_rate": 0.00019594451223097788,
      "loss": 1.1286,
      "step": 1641
    },
    {
      "epoch": 0.11835940315721186,
      "grad_norm": 0.17279403250421443,
      "learning_rate": 0.0001959379281599322,
      "loss": 1.084,
      "step": 1642
    },
    {
      "epoch": 0.11843148561954876,
      "grad_norm": 0.10589860402105457,
      "learning_rate": 0.00019593133885941112,
      "loss": 1.1611,
      "step": 1643
    },
    {
      "epoch": 0.11850356808188568,
      "grad_norm": 0.1293595537969499,
      "learning_rate": 0.00019592474432977386,
      "loss": 1.0447,
      "step": 1644
    },
    {
      "epoch": 0.11857565054422259,
      "grad_norm": 0.1559734868197142,
      "learning_rate": 0.00019591814457137985,
      "loss": 1.1023,
      "step": 1645
    },
    {
      "epoch": 0.11864773300655951,
      "grad_norm": 0.5206925256460665,
      "learning_rate": 0.00019591153958458884,
      "loss": 1.2089,
      "step": 1646
    },
    {
      "epoch": 0.11871981546889641,
      "grad_norm": 0.10593594391794064,
      "learning_rate": 0.00019590492936976092,
      "loss": 1.1841,
      "step": 1647
    },
    {
      "epoch": 0.11879189793123333,
      "grad_norm": 0.19713487115265738,
      "learning_rate": 0.00019589831392725633,
      "loss": 1.046,
      "step": 1648
    },
    {
      "epoch": 0.11886398039357024,
      "grad_norm": 0.2099833978732743,
      "learning_rate": 0.00019589169325743574,
      "loss": 1.1468,
      "step": 1649
    },
    {
      "epoch": 0.11893606285590716,
      "grad_norm": 0.12772380042073997,
      "learning_rate": 0.00019588506736065994,
      "loss": 1.144,
      "step": 1650
    },
    {
      "epoch": 0.11900814531824407,
      "grad_norm": 0.25219181777360294,
      "learning_rate": 0.00019587843623729016,
      "loss": 1.1398,
      "step": 1651
    },
    {
      "epoch": 0.11908022778058099,
      "grad_norm": 0.1562721999769911,
      "learning_rate": 0.0001958717998876879,
      "loss": 1.1525,
      "step": 1652
    },
    {
      "epoch": 0.11915231024291789,
      "grad_norm": 0.1823391601760601,
      "learning_rate": 0.0001958651583122148,
      "loss": 1.0259,
      "step": 1653
    },
    {
      "epoch": 0.11922439270525481,
      "grad_norm": 0.2033024492564876,
      "learning_rate": 0.00019585851151123298,
      "loss": 1.0879,
      "step": 1654
    },
    {
      "epoch": 0.11929647516759173,
      "grad_norm": 0.13924379357944225,
      "learning_rate": 0.0001958518594851047,
      "loss": 1.0617,
      "step": 1655
    },
    {
      "epoch": 0.11936855762992864,
      "grad_norm": 0.17779428866527677,
      "learning_rate": 0.00019584520223419255,
      "loss": 1.1897,
      "step": 1656
    },
    {
      "epoch": 0.11944064009226556,
      "grad_norm": 0.1367724818505277,
      "learning_rate": 0.0001958385397588594,
      "loss": 1.1746,
      "step": 1657
    },
    {
      "epoch": 0.11951272255460246,
      "grad_norm": 0.12165721593393941,
      "learning_rate": 0.00019583187205946846,
      "loss": 1.0673,
      "step": 1658
    },
    {
      "epoch": 0.11958480501693938,
      "grad_norm": 0.2132438330776148,
      "learning_rate": 0.00019582519913638317,
      "loss": 1.1467,
      "step": 1659
    },
    {
      "epoch": 0.11965688747927629,
      "grad_norm": 0.11638211289906492,
      "learning_rate": 0.00019581852098996726,
      "loss": 1.1183,
      "step": 1660
    },
    {
      "epoch": 0.11972896994161321,
      "grad_norm": 0.1435866020353608,
      "learning_rate": 0.0001958118376205847,
      "loss": 1.1066,
      "step": 1661
    },
    {
      "epoch": 0.11980105240395011,
      "grad_norm": 0.21319966758739942,
      "learning_rate": 0.00019580514902859986,
      "loss": 1.086,
      "step": 1662
    },
    {
      "epoch": 0.11987313486628703,
      "grad_norm": 0.3031008867620505,
      "learning_rate": 0.00019579845521437727,
      "loss": 1.1908,
      "step": 1663
    },
    {
      "epoch": 0.11994521732862394,
      "grad_norm": 0.17373139200080312,
      "learning_rate": 0.00019579175617828187,
      "loss": 1.0746,
      "step": 1664
    },
    {
      "epoch": 0.12001729979096086,
      "grad_norm": 0.10958986792200047,
      "learning_rate": 0.00019578505192067874,
      "loss": 1.1338,
      "step": 1665
    },
    {
      "epoch": 0.12008938225329777,
      "grad_norm": 0.14136427869918078,
      "learning_rate": 0.0001957783424419334,
      "loss": 1.0586,
      "step": 1666
    },
    {
      "epoch": 0.12016146471563469,
      "grad_norm": 0.12305313284912027,
      "learning_rate": 0.00019577162774241154,
      "loss": 1.1044,
      "step": 1667
    },
    {
      "epoch": 0.1202335471779716,
      "grad_norm": 0.187791285580268,
      "learning_rate": 0.00019576490782247916,
      "loss": 1.1687,
      "step": 1668
    },
    {
      "epoch": 0.12030562964030851,
      "grad_norm": 0.2920558442890481,
      "learning_rate": 0.00019575818268250257,
      "loss": 1.2125,
      "step": 1669
    },
    {
      "epoch": 0.12037771210264543,
      "grad_norm": 0.12567812740639986,
      "learning_rate": 0.00019575145232284832,
      "loss": 1.1632,
      "step": 1670
    },
    {
      "epoch": 0.12044979456498234,
      "grad_norm": 0.1131786584853098,
      "learning_rate": 0.00019574471674388332,
      "loss": 1.1577,
      "step": 1671
    },
    {
      "epoch": 0.12052187702731926,
      "grad_norm": 0.13519123546849487,
      "learning_rate": 0.00019573797594597472,
      "loss": 1.146,
      "step": 1672
    },
    {
      "epoch": 0.12059395948965616,
      "grad_norm": 0.12280039155638217,
      "learning_rate": 0.0001957312299294899,
      "loss": 1.1725,
      "step": 1673
    },
    {
      "epoch": 0.12066604195199308,
      "grad_norm": 0.2602532818356042,
      "learning_rate": 0.00019572447869479665,
      "loss": 1.1338,
      "step": 1674
    },
    {
      "epoch": 0.12073812441432999,
      "grad_norm": 0.1219034192053005,
      "learning_rate": 0.00019571772224226291,
      "loss": 1.2693,
      "step": 1675
    },
    {
      "epoch": 0.12081020687666691,
      "grad_norm": 0.14269228012383653,
      "learning_rate": 0.000195710960572257,
      "loss": 1.071,
      "step": 1676
    },
    {
      "epoch": 0.12088228933900382,
      "grad_norm": 0.5521767803877436,
      "learning_rate": 0.0001957041936851475,
      "loss": 1.0223,
      "step": 1677
    },
    {
      "epoch": 0.12095437180134073,
      "grad_norm": 0.1641500661650213,
      "learning_rate": 0.00019569742158130323,
      "loss": 1.0807,
      "step": 1678
    },
    {
      "epoch": 0.12102645426367764,
      "grad_norm": 0.11995624293673593,
      "learning_rate": 0.00019569064426109333,
      "loss": 1.0878,
      "step": 1679
    },
    {
      "epoch": 0.12109853672601456,
      "grad_norm": 0.14146315334159876,
      "learning_rate": 0.0001956838617248873,
      "loss": 1.0597,
      "step": 1680
    },
    {
      "epoch": 0.12117061918835148,
      "grad_norm": 0.10047793635245361,
      "learning_rate": 0.00019567707397305474,
      "loss": 1.1447,
      "step": 1681
    },
    {
      "epoch": 0.12124270165068839,
      "grad_norm": 0.1721549602728933,
      "learning_rate": 0.0001956702810059657,
      "loss": 1.1071,
      "step": 1682
    },
    {
      "epoch": 0.1213147841130253,
      "grad_norm": 0.1906386900941106,
      "learning_rate": 0.0001956634828239905,
      "loss": 1.0887,
      "step": 1683
    },
    {
      "epoch": 0.12138686657536221,
      "grad_norm": 0.13876874166193168,
      "learning_rate": 0.0001956566794274996,
      "loss": 1.1387,
      "step": 1684
    },
    {
      "epoch": 0.12145894903769913,
      "grad_norm": 0.15078855420030757,
      "learning_rate": 0.00019564987081686395,
      "loss": 1.1412,
      "step": 1685
    },
    {
      "epoch": 0.12153103150003604,
      "grad_norm": 0.1869749492261553,
      "learning_rate": 0.0001956430569924546,
      "loss": 1.1279,
      "step": 1686
    },
    {
      "epoch": 0.12160311396237296,
      "grad_norm": 0.11243413725732615,
      "learning_rate": 0.00019563623795464303,
      "loss": 1.0907,
      "step": 1687
    },
    {
      "epoch": 0.12167519642470986,
      "grad_norm": 0.19601390936897317,
      "learning_rate": 0.00019562941370380088,
      "loss": 1.1611,
      "step": 1688
    },
    {
      "epoch": 0.12174727888704678,
      "grad_norm": 1.4382416931394713,
      "learning_rate": 0.00019562258424030016,
      "loss": 1.1291,
      "step": 1689
    },
    {
      "epoch": 0.12181936134938369,
      "grad_norm": 0.11399459555241877,
      "learning_rate": 0.00019561574956451314,
      "loss": 1.1444,
      "step": 1690
    },
    {
      "epoch": 0.12189144381172061,
      "grad_norm": 0.11754522337838935,
      "learning_rate": 0.00019560890967681237,
      "loss": 1.1351,
      "step": 1691
    },
    {
      "epoch": 0.12196352627405752,
      "grad_norm": 0.2927421651770483,
      "learning_rate": 0.00019560206457757066,
      "loss": 1.0804,
      "step": 1692
    },
    {
      "epoch": 0.12203560873639444,
      "grad_norm": 1.048715697768011,
      "learning_rate": 0.00019559521426716118,
      "loss": 1.1291,
      "step": 1693
    },
    {
      "epoch": 0.12210769119873136,
      "grad_norm": 0.14313892687729984,
      "learning_rate": 0.00019558835874595727,
      "loss": 1.0723,
      "step": 1694
    },
    {
      "epoch": 0.12217977366106826,
      "grad_norm": 0.10796161426031022,
      "learning_rate": 0.00019558149801433266,
      "loss": 1.0835,
      "step": 1695
    },
    {
      "epoch": 0.12225185612340518,
      "grad_norm": 0.3531640525179138,
      "learning_rate": 0.00019557463207266134,
      "loss": 1.2081,
      "step": 1696
    },
    {
      "epoch": 0.12232393858574209,
      "grad_norm": 5.167863454183616,
      "learning_rate": 0.00019556776092131753,
      "loss": 1.121,
      "step": 1697
    },
    {
      "epoch": 0.122396021048079,
      "grad_norm": 0.11619799006349994,
      "learning_rate": 0.00019556088456067574,
      "loss": 1.0691,
      "step": 1698
    },
    {
      "epoch": 0.12246810351041591,
      "grad_norm": 0.6987999943536077,
      "learning_rate": 0.00019555400299111085,
      "loss": 1.0375,
      "step": 1699
    },
    {
      "epoch": 0.12254018597275283,
      "grad_norm": 0.16512008372445486,
      "learning_rate": 0.00019554711621299795,
      "loss": 1.1296,
      "step": 1700
    },
    {
      "epoch": 0.12261226843508974,
      "grad_norm": 0.1743647646347656,
      "learning_rate": 0.00019554022422671243,
      "loss": 1.0436,
      "step": 1701
    },
    {
      "epoch": 0.12268435089742666,
      "grad_norm": 1.0132016645929907,
      "learning_rate": 0.00019553332703262997,
      "loss": 1.1078,
      "step": 1702
    },
    {
      "epoch": 0.12275643335976356,
      "grad_norm": 0.272135741479582,
      "learning_rate": 0.0001955264246311265,
      "loss": 1.1543,
      "step": 1703
    },
    {
      "epoch": 0.12282851582210048,
      "grad_norm": 0.24324981582996671,
      "learning_rate": 0.0001955195170225783,
      "loss": 1.1079,
      "step": 1704
    },
    {
      "epoch": 0.12290059828443739,
      "grad_norm": 0.23487632821556118,
      "learning_rate": 0.00019551260420736189,
      "loss": 1.2091,
      "step": 1705
    },
    {
      "epoch": 0.12297268074677431,
      "grad_norm": 0.17142470540809138,
      "learning_rate": 0.00019550568618585402,
      "loss": 1.1425,
      "step": 1706
    },
    {
      "epoch": 0.12304476320911123,
      "grad_norm": 0.16220525702540114,
      "learning_rate": 0.00019549876295843187,
      "loss": 1.1045,
      "step": 1707
    },
    {
      "epoch": 0.12311684567144814,
      "grad_norm": 0.12696821454800525,
      "learning_rate": 0.0001954918345254728,
      "loss": 1.0827,
      "step": 1708
    },
    {
      "epoch": 0.12318892813378506,
      "grad_norm": 0.3862440725294241,
      "learning_rate": 0.00019548490088735444,
      "loss": 1.0985,
      "step": 1709
    },
    {
      "epoch": 0.12326101059612196,
      "grad_norm": 0.16435452853658133,
      "learning_rate": 0.00019547796204445473,
      "loss": 1.1451,
      "step": 1710
    },
    {
      "epoch": 0.12333309305845888,
      "grad_norm": 0.18609515659120923,
      "learning_rate": 0.00019547101799715196,
      "loss": 1.1817,
      "step": 1711
    },
    {
      "epoch": 0.12340517552079579,
      "grad_norm": 0.37471316906011826,
      "learning_rate": 0.00019546406874582456,
      "loss": 1.1243,
      "step": 1712
    },
    {
      "epoch": 0.12347725798313271,
      "grad_norm": 0.1896190820103745,
      "learning_rate": 0.00019545711429085137,
      "loss": 1.1376,
      "step": 1713
    },
    {
      "epoch": 0.12354934044546961,
      "grad_norm": 0.14084416477511383,
      "learning_rate": 0.00019545015463261146,
      "loss": 1.0816,
      "step": 1714
    },
    {
      "epoch": 0.12362142290780653,
      "grad_norm": 0.4060857882753813,
      "learning_rate": 0.00019544318977148426,
      "loss": 1.0879,
      "step": 1715
    },
    {
      "epoch": 0.12369350537014344,
      "grad_norm": 0.4487754105387071,
      "learning_rate": 0.00019543621970784928,
      "loss": 1.1341,
      "step": 1716
    },
    {
      "epoch": 0.12376558783248036,
      "grad_norm": 2.151532931694754,
      "learning_rate": 0.00019542924444208658,
      "loss": 1.075,
      "step": 1717
    },
    {
      "epoch": 0.12383767029481726,
      "grad_norm": 0.16033772566171817,
      "learning_rate": 0.00019542226397457633,
      "loss": 1.1281,
      "step": 1718
    },
    {
      "epoch": 0.12390975275715418,
      "grad_norm": 0.23436556298533742,
      "learning_rate": 0.000195415278305699,
      "loss": 1.1159,
      "step": 1719
    },
    {
      "epoch": 0.1239818352194911,
      "grad_norm": 2.4005923342523197,
      "learning_rate": 0.00019540828743583538,
      "loss": 1.1271,
      "step": 1720
    },
    {
      "epoch": 0.12405391768182801,
      "grad_norm": 0.2710835673638073,
      "learning_rate": 0.0001954012913653666,
      "loss": 1.19,
      "step": 1721
    },
    {
      "epoch": 0.12412600014416493,
      "grad_norm": 0.1339772699725737,
      "learning_rate": 0.0001953942900946739,
      "loss": 1.1311,
      "step": 1722
    },
    {
      "epoch": 0.12419808260650184,
      "grad_norm": 0.1306082281133291,
      "learning_rate": 0.00019538728362413895,
      "loss": 1.1205,
      "step": 1723
    },
    {
      "epoch": 0.12427016506883876,
      "grad_norm": 0.20594870592158526,
      "learning_rate": 0.00019538027195414374,
      "loss": 1.1512,
      "step": 1724
    },
    {
      "epoch": 0.12434224753117566,
      "grad_norm": 0.5574358400780234,
      "learning_rate": 0.00019537325508507038,
      "loss": 1.0872,
      "step": 1725
    },
    {
      "epoch": 0.12441432999351258,
      "grad_norm": 0.2789448818442867,
      "learning_rate": 0.0001953662330173014,
      "loss": 1.0448,
      "step": 1726
    },
    {
      "epoch": 0.12448641245584949,
      "grad_norm": 0.12016197285231194,
      "learning_rate": 0.00019535920575121957,
      "loss": 1.0022,
      "step": 1727
    },
    {
      "epoch": 0.12455849491818641,
      "grad_norm": 0.35836218510887713,
      "learning_rate": 0.0001953521732872079,
      "loss": 1.0503,
      "step": 1728
    },
    {
      "epoch": 0.12463057738052331,
      "grad_norm": 0.4681045374643605,
      "learning_rate": 0.00019534513562564972,
      "loss": 1.1353,
      "step": 1729
    },
    {
      "epoch": 0.12470265984286023,
      "grad_norm": 0.18912823959807007,
      "learning_rate": 0.0001953380927669287,
      "loss": 1.0981,
      "step": 1730
    },
    {
      "epoch": 0.12477474230519714,
      "grad_norm": 0.3568632650748411,
      "learning_rate": 0.00019533104471142866,
      "loss": 1.1562,
      "step": 1731
    },
    {
      "epoch": 0.12484682476753406,
      "grad_norm": 1.1876859036573784,
      "learning_rate": 0.00019532399145953387,
      "loss": 1.0264,
      "step": 1732
    },
    {
      "epoch": 0.12491890722987098,
      "grad_norm": 0.3113699161796197,
      "learning_rate": 0.00019531693301162874,
      "loss": 1.0971,
      "step": 1733
    },
    {
      "epoch": 0.12499098969220789,
      "grad_norm": 0.11097858683505142,
      "learning_rate": 0.00019530986936809803,
      "loss": 1.1279,
      "step": 1734
    },
    {
      "epoch": 0.1250630721545448,
      "grad_norm": 0.5291283878505998,
      "learning_rate": 0.00019530280052932678,
      "loss": 1.1275,
      "step": 1735
    },
    {
      "epoch": 0.1251351546168817,
      "grad_norm": 0.9747001340947793,
      "learning_rate": 0.00019529572649570027,
      "loss": 1.1519,
      "step": 1736
    },
    {
      "epoch": 0.12520723707921863,
      "grad_norm": 0.1196893815777936,
      "learning_rate": 0.00019528864726760416,
      "loss": 1.1571,
      "step": 1737
    },
    {
      "epoch": 0.12527931954155555,
      "grad_norm": 0.16183843485892202,
      "learning_rate": 0.00019528156284542427,
      "loss": 1.1658,
      "step": 1738
    },
    {
      "epoch": 0.12535140200389244,
      "grad_norm": 0.16095033744140289,
      "learning_rate": 0.0001952744732295468,
      "loss": 1.1542,
      "step": 1739
    },
    {
      "epoch": 0.12542348446622936,
      "grad_norm": 0.14932602413985266,
      "learning_rate": 0.0001952673784203582,
      "loss": 1.0802,
      "step": 1740
    },
    {
      "epoch": 0.12549556692856628,
      "grad_norm": 0.1420828338370416,
      "learning_rate": 0.00019526027841824522,
      "loss": 1.0594,
      "step": 1741
    },
    {
      "epoch": 0.1255676493909032,
      "grad_norm": 1.2428767559348177,
      "learning_rate": 0.0001952531732235948,
      "loss": 1.115,
      "step": 1742
    },
    {
      "epoch": 0.1256397318532401,
      "grad_norm": 0.8190898953132225,
      "learning_rate": 0.00019524606283679427,
      "loss": 1.2131,
      "step": 1743
    },
    {
      "epoch": 0.12571181431557701,
      "grad_norm": 0.13289489601359175,
      "learning_rate": 0.00019523894725823125,
      "loss": 1.1246,
      "step": 1744
    },
    {
      "epoch": 0.12578389677791393,
      "grad_norm": 0.19445717540279642,
      "learning_rate": 0.00019523182648829358,
      "loss": 1.1123,
      "step": 1745
    },
    {
      "epoch": 0.12585597924025085,
      "grad_norm": 0.130806549504771,
      "learning_rate": 0.00019522470052736935,
      "loss": 1.1421,
      "step": 1746
    },
    {
      "epoch": 0.12592806170258777,
      "grad_norm": 0.13124016895030957,
      "learning_rate": 0.0001952175693758471,
      "loss": 1.1222,
      "step": 1747
    },
    {
      "epoch": 0.12600014416492467,
      "grad_norm": 0.11225668891199048,
      "learning_rate": 0.00019521043303411542,
      "loss": 1.0115,
      "step": 1748
    },
    {
      "epoch": 0.12607222662726159,
      "grad_norm": 0.1379275875341082,
      "learning_rate": 0.00019520329150256337,
      "loss": 1.0073,
      "step": 1749
    },
    {
      "epoch": 0.1261443090895985,
      "grad_norm": 0.10969623747695102,
      "learning_rate": 0.00019519614478158027,
      "loss": 1.1343,
      "step": 1750
    },
    {
      "epoch": 0.12621639155193543,
      "grad_norm": 0.18661380909947667,
      "learning_rate": 0.00019518899287155556,
      "loss": 1.0697,
      "step": 1751
    },
    {
      "epoch": 0.12628847401427232,
      "grad_norm": 0.5012035165406146,
      "learning_rate": 0.0001951818357728792,
      "loss": 1.0966,
      "step": 1752
    },
    {
      "epoch": 0.12636055647660924,
      "grad_norm": 0.2883044295605109,
      "learning_rate": 0.00019517467348594125,
      "loss": 1.1038,
      "step": 1753
    },
    {
      "epoch": 0.12643263893894616,
      "grad_norm": 0.2635131924244614,
      "learning_rate": 0.00019516750601113216,
      "loss": 1.1243,
      "step": 1754
    },
    {
      "epoch": 0.12650472140128308,
      "grad_norm": 0.16005894220230854,
      "learning_rate": 0.00019516033334884258,
      "loss": 1.1244,
      "step": 1755
    },
    {
      "epoch": 0.12657680386361997,
      "grad_norm": 0.13845129906023293,
      "learning_rate": 0.0001951531554994635,
      "loss": 1.05,
      "step": 1756
    },
    {
      "epoch": 0.1266488863259569,
      "grad_norm": 0.14454059039399525,
      "learning_rate": 0.00019514597246338621,
      "loss": 1.1588,
      "step": 1757
    },
    {
      "epoch": 0.1267209687882938,
      "grad_norm": 0.12408829342685171,
      "learning_rate": 0.0001951387842410022,
      "loss": 1.1129,
      "step": 1758
    },
    {
      "epoch": 0.12679305125063073,
      "grad_norm": 0.09897743120153593,
      "learning_rate": 0.00019513159083270332,
      "loss": 1.078,
      "step": 1759
    },
    {
      "epoch": 0.12686513371296765,
      "grad_norm": 0.3773138809605222,
      "learning_rate": 0.00019512439223888167,
      "loss": 1.1507,
      "step": 1760
    },
    {
      "epoch": 0.12693721617530454,
      "grad_norm": 0.2883556969302062,
      "learning_rate": 0.00019511718845992962,
      "loss": 1.1507,
      "step": 1761
    },
    {
      "epoch": 0.12700929863764146,
      "grad_norm": 1.4369810874254594,
      "learning_rate": 0.00019510997949623987,
      "loss": 1.1048,
      "step": 1762
    },
    {
      "epoch": 0.12708138109997838,
      "grad_norm": 0.10857552489462355,
      "learning_rate": 0.00019510276534820536,
      "loss": 1.0883,
      "step": 1763
    },
    {
      "epoch": 0.1271534635623153,
      "grad_norm": 0.14418105788903046,
      "learning_rate": 0.0001950955460162193,
      "loss": 1.2043,
      "step": 1764
    },
    {
      "epoch": 0.1272255460246522,
      "grad_norm": 0.2245918323316876,
      "learning_rate": 0.00019508832150067526,
      "loss": 1.0879,
      "step": 1765
    },
    {
      "epoch": 0.1272976284869891,
      "grad_norm": 0.1685336760938127,
      "learning_rate": 0.000195081091801967,
      "loss": 1.2079,
      "step": 1766
    },
    {
      "epoch": 0.12736971094932603,
      "grad_norm": 0.13921489454511052,
      "learning_rate": 0.00019507385692048862,
      "loss": 1.1095,
      "step": 1767
    },
    {
      "epoch": 0.12744179341166295,
      "grad_norm": 0.16306334080970894,
      "learning_rate": 0.0001950666168566345,
      "loss": 1.0736,
      "step": 1768
    },
    {
      "epoch": 0.12751387587399984,
      "grad_norm": 0.19659953147219825,
      "learning_rate": 0.00019505937161079927,
      "loss": 1.0463,
      "step": 1769
    },
    {
      "epoch": 0.12758595833633676,
      "grad_norm": 0.17827365754331143,
      "learning_rate": 0.00019505212118337784,
      "loss": 1.1601,
      "step": 1770
    },
    {
      "epoch": 0.12765804079867368,
      "grad_norm": 0.11889906553398251,
      "learning_rate": 0.00019504486557476544,
      "loss": 1.1679,
      "step": 1771
    },
    {
      "epoch": 0.1277301232610106,
      "grad_norm": 0.11662066630000718,
      "learning_rate": 0.0001950376047853576,
      "loss": 1.2084,
      "step": 1772
    },
    {
      "epoch": 0.12780220572334752,
      "grad_norm": 0.3305643578253237,
      "learning_rate": 0.00019503033881555006,
      "loss": 1.0472,
      "step": 1773
    },
    {
      "epoch": 0.12787428818568441,
      "grad_norm": 0.10487552523228616,
      "learning_rate": 0.00019502306766573888,
      "loss": 1.0674,
      "step": 1774
    },
    {
      "epoch": 0.12794637064802133,
      "grad_norm": 0.11449313508522942,
      "learning_rate": 0.0001950157913363204,
      "loss": 1.0861,
      "step": 1775
    },
    {
      "epoch": 0.12801845311035825,
      "grad_norm": 0.48978097298494067,
      "learning_rate": 0.00019500850982769128,
      "loss": 1.0861,
      "step": 1776
    },
    {
      "epoch": 0.12809053557269517,
      "grad_norm": 0.13668167614396876,
      "learning_rate": 0.0001950012231402484,
      "loss": 1.1732,
      "step": 1777
    },
    {
      "epoch": 0.12816261803503207,
      "grad_norm": 0.3560818642828099,
      "learning_rate": 0.00019499393127438893,
      "loss": 1.1386,
      "step": 1778
    },
    {
      "epoch": 0.128234700497369,
      "grad_norm": 0.10687160456915637,
      "learning_rate": 0.0001949866342305104,
      "loss": 1.0945,
      "step": 1779
    },
    {
      "epoch": 0.1283067829597059,
      "grad_norm": 0.11825192543857155,
      "learning_rate": 0.0001949793320090105,
      "loss": 1.1317,
      "step": 1780
    },
    {
      "epoch": 0.12837886542204283,
      "grad_norm": 0.11615680597545058,
      "learning_rate": 0.0001949720246102873,
      "loss": 1.0313,
      "step": 1781
    },
    {
      "epoch": 0.12845094788437972,
      "grad_norm": 0.1249951474643181,
      "learning_rate": 0.00019496471203473914,
      "loss": 1.0605,
      "step": 1782
    },
    {
      "epoch": 0.12852303034671664,
      "grad_norm": 0.6163183950497366,
      "learning_rate": 0.00019495739428276458,
      "loss": 1.1465,
      "step": 1783
    },
    {
      "epoch": 0.12859511280905356,
      "grad_norm": 0.11537987050133211,
      "learning_rate": 0.0001949500713547625,
      "loss": 1.0594,
      "step": 1784
    },
    {
      "epoch": 0.12866719527139048,
      "grad_norm": 0.1507303066919531,
      "learning_rate": 0.0001949427432511321,
      "loss": 1.061,
      "step": 1785
    },
    {
      "epoch": 0.1287392777337274,
      "grad_norm": 0.2267277836176224,
      "learning_rate": 0.00019493540997227285,
      "loss": 1.1712,
      "step": 1786
    },
    {
      "epoch": 0.1288113601960643,
      "grad_norm": 0.15713988394248504,
      "learning_rate": 0.00019492807151858438,
      "loss": 1.1147,
      "step": 1787
    },
    {
      "epoch": 0.1288834426584012,
      "grad_norm": 0.12565989423743856,
      "learning_rate": 0.00019492072789046677,
      "loss": 1.1342,
      "step": 1788
    },
    {
      "epoch": 0.12895552512073813,
      "grad_norm": 0.1042154062417135,
      "learning_rate": 0.00019491337908832031,
      "loss": 1.1243,
      "step": 1789
    },
    {
      "epoch": 0.12902760758307505,
      "grad_norm": 0.11859967063972726,
      "learning_rate": 0.0001949060251125456,
      "loss": 1.1721,
      "step": 1790
    },
    {
      "epoch": 0.12909969004541194,
      "grad_norm": 0.23799621510748895,
      "learning_rate": 0.00019489866596354344,
      "loss": 1.1279,
      "step": 1791
    },
    {
      "epoch": 0.12917177250774886,
      "grad_norm": 0.11475894140643335,
      "learning_rate": 0.000194891301641715,
      "loss": 1.0403,
      "step": 1792
    },
    {
      "epoch": 0.12924385497008578,
      "grad_norm": 0.13936324867593822,
      "learning_rate": 0.00019488393214746173,
      "loss": 1.1042,
      "step": 1793
    },
    {
      "epoch": 0.1293159374324227,
      "grad_norm": 0.11506617980209449,
      "learning_rate": 0.00019487655748118526,
      "loss": 1.1977,
      "step": 1794
    },
    {
      "epoch": 0.1293880198947596,
      "grad_norm": 0.11638486736815601,
      "learning_rate": 0.00019486917764328763,
      "loss": 1.0945,
      "step": 1795
    },
    {
      "epoch": 0.1294601023570965,
      "grad_norm": 0.3023230087051109,
      "learning_rate": 0.0001948617926341711,
      "loss": 1.0747,
      "step": 1796
    },
    {
      "epoch": 0.12953218481943343,
      "grad_norm": 0.12009883471419992,
      "learning_rate": 0.0001948544024542382,
      "loss": 1.0669,
      "step": 1797
    },
    {
      "epoch": 0.12960426728177035,
      "grad_norm": 0.11931705047407992,
      "learning_rate": 0.00019484700710389177,
      "loss": 1.0704,
      "step": 1798
    },
    {
      "epoch": 0.12967634974410727,
      "grad_norm": 0.12987465986110883,
      "learning_rate": 0.00019483960658353496,
      "loss": 1.0515,
      "step": 1799
    },
    {
      "epoch": 0.12974843220644416,
      "grad_norm": 0.1232331891663599,
      "learning_rate": 0.0001948322008935711,
      "loss": 1.17,
      "step": 1800
    },
    {
      "epoch": 0.12982051466878108,
      "grad_norm": 0.130134410526344,
      "learning_rate": 0.00019482479003440394,
      "loss": 1.0424,
      "step": 1801
    },
    {
      "epoch": 0.129892597131118,
      "grad_norm": 0.15026377384225043,
      "learning_rate": 0.00019481737400643735,
      "loss": 1.0739,
      "step": 1802
    },
    {
      "epoch": 0.12996467959345492,
      "grad_norm": 0.10692676335775675,
      "learning_rate": 0.00019480995281007565,
      "loss": 1.1801,
      "step": 1803
    },
    {
      "epoch": 0.13003676205579182,
      "grad_norm": 0.5372888738708789,
      "learning_rate": 0.0001948025264457233,
      "loss": 1.0885,
      "step": 1804
    },
    {
      "epoch": 0.13010884451812874,
      "grad_norm": 0.10697834457761028,
      "learning_rate": 0.00019479509491378516,
      "loss": 1.1593,
      "step": 1805
    },
    {
      "epoch": 0.13018092698046566,
      "grad_norm": 0.1266354695762264,
      "learning_rate": 0.00019478765821466623,
      "loss": 1.2444,
      "step": 1806
    },
    {
      "epoch": 0.13025300944280258,
      "grad_norm": 0.18308425631907352,
      "learning_rate": 0.000194780216348772,
      "loss": 1.2031,
      "step": 1807
    },
    {
      "epoch": 0.13032509190513947,
      "grad_norm": 0.15310613840119722,
      "learning_rate": 0.000194772769316508,
      "loss": 1.0108,
      "step": 1808
    },
    {
      "epoch": 0.1303971743674764,
      "grad_norm": 0.12515450881255966,
      "learning_rate": 0.00019476531711828027,
      "loss": 1.1853,
      "step": 1809
    },
    {
      "epoch": 0.1304692568298133,
      "grad_norm": 0.16849665034268788,
      "learning_rate": 0.0001947578597544949,
      "loss": 1.0447,
      "step": 1810
    },
    {
      "epoch": 0.13054133929215023,
      "grad_norm": 0.10790371403824461,
      "learning_rate": 0.00019475039722555848,
      "loss": 1.1307,
      "step": 1811
    },
    {
      "epoch": 0.13061342175448715,
      "grad_norm": 0.10914271010569788,
      "learning_rate": 0.00019474292953187774,
      "loss": 1.155,
      "step": 1812
    },
    {
      "epoch": 0.13068550421682404,
      "grad_norm": 0.13162470849154523,
      "learning_rate": 0.00019473545667385975,
      "loss": 1.116,
      "step": 1813
    },
    {
      "epoch": 0.13075758667916096,
      "grad_norm": 0.12049637644229108,
      "learning_rate": 0.00019472797865191182,
      "loss": 1.0636,
      "step": 1814
    },
    {
      "epoch": 0.13082966914149788,
      "grad_norm": 0.10295961500117286,
      "learning_rate": 0.00019472049546644162,
      "loss": 1.1444,
      "step": 1815
    },
    {
      "epoch": 0.1309017516038348,
      "grad_norm": 0.10707603360922646,
      "learning_rate": 0.00019471300711785703,
      "loss": 1.1388,
      "step": 1816
    },
    {
      "epoch": 0.1309738340661717,
      "grad_norm": 0.0991273283563605,
      "learning_rate": 0.00019470551360656623,
      "loss": 1.1693,
      "step": 1817
    },
    {
      "epoch": 0.1310459165285086,
      "grad_norm": 0.2950150920148713,
      "learning_rate": 0.00019469801493297766,
      "loss": 1.1652,
      "step": 1818
    },
    {
      "epoch": 0.13111799899084553,
      "grad_norm": 0.1095279220849569,
      "learning_rate": 0.00019469051109750006,
      "loss": 1.0383,
      "step": 1819
    },
    {
      "epoch": 0.13119008145318245,
      "grad_norm": 0.34740504813421536,
      "learning_rate": 0.00019468300210054252,
      "loss": 1.1601,
      "step": 1820
    },
    {
      "epoch": 0.13126216391551934,
      "grad_norm": 0.21434138277565493,
      "learning_rate": 0.00019467548794251428,
      "loss": 1.0617,
      "step": 1821
    },
    {
      "epoch": 0.13133424637785626,
      "grad_norm": 0.1161592343896271,
      "learning_rate": 0.00019466796862382496,
      "loss": 1.0703,
      "step": 1822
    },
    {
      "epoch": 0.13140632884019318,
      "grad_norm": 0.1404009510166462,
      "learning_rate": 0.00019466044414488445,
      "loss": 1.1914,
      "step": 1823
    },
    {
      "epoch": 0.1314784113025301,
      "grad_norm": 0.12420230151956603,
      "learning_rate": 0.00019465291450610286,
      "loss": 1.1059,
      "step": 1824
    },
    {
      "epoch": 0.13155049376486702,
      "grad_norm": 0.2547705189313818,
      "learning_rate": 0.00019464537970789066,
      "loss": 1.0217,
      "step": 1825
    },
    {
      "epoch": 0.1316225762272039,
      "grad_norm": 0.10635622800645807,
      "learning_rate": 0.0001946378397506585,
      "loss": 1.0779,
      "step": 1826
    },
    {
      "epoch": 0.13169465868954083,
      "grad_norm": 0.1319058802532492,
      "learning_rate": 0.00019463029463481746,
      "loss": 1.1095,
      "step": 1827
    },
    {
      "epoch": 0.13176674115187775,
      "grad_norm": 0.12725815956205097,
      "learning_rate": 0.00019462274436077875,
      "loss": 1.0618,
      "step": 1828
    },
    {
      "epoch": 0.13183882361421467,
      "grad_norm": 0.11024996033274317,
      "learning_rate": 0.00019461518892895395,
      "loss": 1.2162,
      "step": 1829
    },
    {
      "epoch": 0.13191090607655157,
      "grad_norm": 0.10523940042956992,
      "learning_rate": 0.00019460762833975492,
      "loss": 1.0707,
      "step": 1830
    },
    {
      "epoch": 0.13198298853888848,
      "grad_norm": 0.10072904933686215,
      "learning_rate": 0.00019460006259359376,
      "loss": 1.1346,
      "step": 1831
    },
    {
      "epoch": 0.1320550710012254,
      "grad_norm": 0.12128196357639819,
      "learning_rate": 0.00019459249169088288,
      "loss": 1.14,
      "step": 1832
    },
    {
      "epoch": 0.13212715346356232,
      "grad_norm": 0.09822786372829252,
      "learning_rate": 0.00019458491563203493,
      "loss": 1.0353,
      "step": 1833
    },
    {
      "epoch": 0.13219923592589922,
      "grad_norm": 0.5451229282460326,
      "learning_rate": 0.00019457733441746294,
      "loss": 1.0313,
      "step": 1834
    },
    {
      "epoch": 0.13227131838823614,
      "grad_norm": 0.12156478047649669,
      "learning_rate": 0.00019456974804758006,
      "loss": 1.1355,
      "step": 1835
    },
    {
      "epoch": 0.13234340085057306,
      "grad_norm": 0.3943512119635673,
      "learning_rate": 0.0001945621565227999,
      "loss": 1.0742,
      "step": 1836
    },
    {
      "epoch": 0.13241548331290998,
      "grad_norm": 1.539339571491642,
      "learning_rate": 0.00019455455984353622,
      "loss": 1.0164,
      "step": 1837
    },
    {
      "epoch": 0.1324875657752469,
      "grad_norm": 0.12131117183964338,
      "learning_rate": 0.0001945469580102031,
      "loss": 1.1017,
      "step": 1838
    },
    {
      "epoch": 0.1325596482375838,
      "grad_norm": 0.4357592291017037,
      "learning_rate": 0.00019453935102321497,
      "loss": 1.1092,
      "step": 1839
    },
    {
      "epoch": 0.1326317306999207,
      "grad_norm": 0.13016203097935355,
      "learning_rate": 0.0001945317388829864,
      "loss": 1.1292,
      "step": 1840
    },
    {
      "epoch": 0.13270381316225763,
      "grad_norm": 0.11030571596935676,
      "learning_rate": 0.00019452412158993236,
      "loss": 1.1252,
      "step": 1841
    },
    {
      "epoch": 0.13277589562459455,
      "grad_norm": 0.1215310862984502,
      "learning_rate": 0.00019451649914446807,
      "loss": 1.0835,
      "step": 1842
    },
    {
      "epoch": 0.13284797808693144,
      "grad_norm": 0.2219166223419539,
      "learning_rate": 0.00019450887154700902,
      "loss": 1.0976,
      "step": 1843
    },
    {
      "epoch": 0.13292006054926836,
      "grad_norm": 0.15547905277888494,
      "learning_rate": 0.00019450123879797095,
      "loss": 1.0651,
      "step": 1844
    },
    {
      "epoch": 0.13299214301160528,
      "grad_norm": 1.2664320462116763,
      "learning_rate": 0.00019449360089776995,
      "loss": 1.0527,
      "step": 1845
    },
    {
      "epoch": 0.1330642254739422,
      "grad_norm": 0.12336279907702628,
      "learning_rate": 0.00019448595784682228,
      "loss": 1.0944,
      "step": 1846
    },
    {
      "epoch": 0.1331363079362791,
      "grad_norm": 0.36404906038414386,
      "learning_rate": 0.00019447830964554468,
      "loss": 1.2084,
      "step": 1847
    },
    {
      "epoch": 0.133208390398616,
      "grad_norm": 0.15585657470646214,
      "learning_rate": 0.00019447065629435396,
      "loss": 1.1228,
      "step": 1848
    },
    {
      "epoch": 0.13328047286095293,
      "grad_norm": 0.14289024538596995,
      "learning_rate": 0.0001944629977936673,
      "loss": 1.1049,
      "step": 1849
    },
    {
      "epoch": 0.13335255532328985,
      "grad_norm": 0.18368602719380334,
      "learning_rate": 0.00019445533414390214,
      "loss": 1.1469,
      "step": 1850
    },
    {
      "epoch": 0.13342463778562677,
      "grad_norm": 0.36712223870150973,
      "learning_rate": 0.00019444766534547632,
      "loss": 1.0813,
      "step": 1851
    },
    {
      "epoch": 0.13349672024796366,
      "grad_norm": 0.12131434110724529,
      "learning_rate": 0.00019443999139880775,
      "loss": 1.1724,
      "step": 1852
    },
    {
      "epoch": 0.13356880271030058,
      "grad_norm": 0.446943928569457,
      "learning_rate": 0.00019443231230431476,
      "loss": 1.0649,
      "step": 1853
    },
    {
      "epoch": 0.1336408851726375,
      "grad_norm": 0.12631192223328924,
      "learning_rate": 0.00019442462806241594,
      "loss": 1.029,
      "step": 1854
    },
    {
      "epoch": 0.13371296763497442,
      "grad_norm": 0.1912018588760802,
      "learning_rate": 0.00019441693867353012,
      "loss": 1.2228,
      "step": 1855
    },
    {
      "epoch": 0.13378505009731131,
      "grad_norm": 0.1743004512945079,
      "learning_rate": 0.0001944092441380765,
      "loss": 1.0909,
      "step": 1856
    },
    {
      "epoch": 0.13385713255964823,
      "grad_norm": 0.13951070409041047,
      "learning_rate": 0.00019440154445647445,
      "loss": 1.0733,
      "step": 1857
    },
    {
      "epoch": 0.13392921502198515,
      "grad_norm": 0.1298156208401925,
      "learning_rate": 0.00019439383962914367,
      "loss": 1.0594,
      "step": 1858
    },
    {
      "epoch": 0.13400129748432207,
      "grad_norm": 0.12224019489172396,
      "learning_rate": 0.0001943861296565042,
      "loss": 1.0778,
      "step": 1859
    },
    {
      "epoch": 0.13407337994665897,
      "grad_norm": 0.10182321343372612,
      "learning_rate": 0.00019437841453897624,
      "loss": 1.108,
      "step": 1860
    },
    {
      "epoch": 0.13414546240899589,
      "grad_norm": 0.16778997419317065,
      "learning_rate": 0.00019437069427698033,
      "loss": 1.1211,
      "step": 1861
    },
    {
      "epoch": 0.1342175448713328,
      "grad_norm": 0.14599014564203816,
      "learning_rate": 0.00019436296887093737,
      "loss": 1.0892,
      "step": 1862
    },
    {
      "epoch": 0.13428962733366973,
      "grad_norm": 0.29163432730860733,
      "learning_rate": 0.00019435523832126835,
      "loss": 1.1073,
      "step": 1863
    },
    {
      "epoch": 0.13436170979600665,
      "grad_norm": 0.12440551177987419,
      "learning_rate": 0.00019434750262839476,
      "loss": 1.1292,
      "step": 1864
    },
    {
      "epoch": 0.13443379225834354,
      "grad_norm": 0.17805573948610737,
      "learning_rate": 0.0001943397617927382,
      "loss": 1.0974,
      "step": 1865
    },
    {
      "epoch": 0.13450587472068046,
      "grad_norm": 0.11465047512944647,
      "learning_rate": 0.00019433201581472065,
      "loss": 1.1118,
      "step": 1866
    },
    {
      "epoch": 0.13457795718301738,
      "grad_norm": 0.2503168547993103,
      "learning_rate": 0.0001943242646947643,
      "loss": 1.1687,
      "step": 1867
    },
    {
      "epoch": 0.1346500396453543,
      "grad_norm": 0.28975608430650135,
      "learning_rate": 0.00019431650843329169,
      "loss": 1.0628,
      "step": 1868
    },
    {
      "epoch": 0.1347221221076912,
      "grad_norm": 0.10926397525024546,
      "learning_rate": 0.00019430874703072558,
      "loss": 1.0984,
      "step": 1869
    },
    {
      "epoch": 0.1347942045700281,
      "grad_norm": 0.11980345347227944,
      "learning_rate": 0.00019430098048748906,
      "loss": 1.1417,
      "step": 1870
    },
    {
      "epoch": 0.13486628703236503,
      "grad_norm": 0.10333946250388513,
      "learning_rate": 0.00019429320880400545,
      "loss": 1.1252,
      "step": 1871
    },
    {
      "epoch": 0.13493836949470195,
      "grad_norm": 0.33554564844828794,
      "learning_rate": 0.0001942854319806984,
      "loss": 1.1361,
      "step": 1872
    },
    {
      "epoch": 0.13501045195703884,
      "grad_norm": 0.10699317276165816,
      "learning_rate": 0.0001942776500179918,
      "loss": 1.072,
      "step": 1873
    },
    {
      "epoch": 0.13508253441937576,
      "grad_norm": 0.11392959696725143,
      "learning_rate": 0.00019426986291630984,
      "loss": 1.099,
      "step": 1874
    },
    {
      "epoch": 0.13515461688171268,
      "grad_norm": 0.14847669338163771,
      "learning_rate": 0.00019426207067607697,
      "loss": 1.0887,
      "step": 1875
    },
    {
      "epoch": 0.1352266993440496,
      "grad_norm": 2.1778308471282624,
      "learning_rate": 0.000194254273297718,
      "loss": 1.1799,
      "step": 1876
    },
    {
      "epoch": 0.13529878180638652,
      "grad_norm": 0.1911921369377726,
      "learning_rate": 0.0001942464707816579,
      "loss": 1.0716,
      "step": 1877
    },
    {
      "epoch": 0.1353708642687234,
      "grad_norm": 0.11177444227472913,
      "learning_rate": 0.00019423866312832195,
      "loss": 0.9512,
      "step": 1878
    },
    {
      "epoch": 0.13544294673106033,
      "grad_norm": 15.873775308253395,
      "learning_rate": 0.0001942308503381358,
      "loss": 1.1446,
      "step": 1879
    },
    {
      "epoch": 0.13551502919339725,
      "grad_norm": 1.131764410479066,
      "learning_rate": 0.00019422303241152533,
      "loss": 1.0364,
      "step": 1880
    },
    {
      "epoch": 0.13558711165573417,
      "grad_norm": 0.2137430607858249,
      "learning_rate": 0.0001942152093489166,
      "loss": 1.0778,
      "step": 1881
    },
    {
      "epoch": 0.13565919411807106,
      "grad_norm": 0.33689465494843485,
      "learning_rate": 0.00019420738115073611,
      "loss": 1.1618,
      "step": 1882
    },
    {
      "epoch": 0.13573127658040798,
      "grad_norm": 0.3894473447442692,
      "learning_rate": 0.00019419954781741054,
      "loss": 1.1647,
      "step": 1883
    },
    {
      "epoch": 0.1358033590427449,
      "grad_norm": 0.4984751490914845,
      "learning_rate": 0.00019419170934936687,
      "loss": 1.1747,
      "step": 1884
    },
    {
      "epoch": 0.13587544150508182,
      "grad_norm": 0.9936819784696916,
      "learning_rate": 0.0001941838657470324,
      "loss": 1.1784,
      "step": 1885
    },
    {
      "epoch": 0.13594752396741872,
      "grad_norm": 0.1618720080932236,
      "learning_rate": 0.00019417601701083464,
      "loss": 1.0845,
      "step": 1886
    },
    {
      "epoch": 0.13601960642975564,
      "grad_norm": 2.015079832624496,
      "learning_rate": 0.00019416816314120144,
      "loss": 1.1326,
      "step": 1887
    },
    {
      "epoch": 0.13609168889209255,
      "grad_norm": 1.5067214337848793,
      "learning_rate": 0.00019416030413856088,
      "loss": 1.0779,
      "step": 1888
    },
    {
      "epoch": 0.13616377135442947,
      "grad_norm": 0.5264784405046443,
      "learning_rate": 0.0001941524400033414,
      "loss": 1.0081,
      "step": 1889
    },
    {
      "epoch": 0.1362358538167664,
      "grad_norm": 0.5721350563701189,
      "learning_rate": 0.00019414457073597158,
      "loss": 1.1768,
      "step": 1890
    },
    {
      "epoch": 0.1363079362791033,
      "grad_norm": 0.3089513799981377,
      "learning_rate": 0.00019413669633688044,
      "loss": 1.0715,
      "step": 1891
    },
    {
      "epoch": 0.1363800187414402,
      "grad_norm": 0.2808293874437253,
      "learning_rate": 0.00019412881680649715,
      "loss": 1.1116,
      "step": 1892
    },
    {
      "epoch": 0.13645210120377713,
      "grad_norm": 0.22425939184401594,
      "learning_rate": 0.00019412093214525128,
      "loss": 1.1249,
      "step": 1893
    },
    {
      "epoch": 0.13652418366611405,
      "grad_norm": 0.21658285318117257,
      "learning_rate": 0.00019411304235357254,
      "loss": 1.2034,
      "step": 1894
    },
    {
      "epoch": 0.13659626612845094,
      "grad_norm": 0.4013696908362657,
      "learning_rate": 0.00019410514743189108,
      "loss": 1.0949,
      "step": 1895
    },
    {
      "epoch": 0.13666834859078786,
      "grad_norm": 0.38384932590456683,
      "learning_rate": 0.00019409724738063714,
      "loss": 1.1697,
      "step": 1896
    },
    {
      "epoch": 0.13674043105312478,
      "grad_norm": 0.26971721160541784,
      "learning_rate": 0.00019408934220024144,
      "loss": 1.1669,
      "step": 1897
    },
    {
      "epoch": 0.1368125135154617,
      "grad_norm": 1.6258944301640539,
      "learning_rate": 0.00019408143189113484,
      "loss": 1.1769,
      "step": 1898
    },
    {
      "epoch": 0.1368845959777986,
      "grad_norm": 0.13804909798825418,
      "learning_rate": 0.0001940735164537485,
      "loss": 1.1284,
      "step": 1899
    },
    {
      "epoch": 0.1369566784401355,
      "grad_norm": 0.38310988103233645,
      "learning_rate": 0.00019406559588851395,
      "loss": 1.111,
      "step": 1900
    },
    {
      "epoch": 0.13702876090247243,
      "grad_norm": 0.6531629281136301,
      "learning_rate": 0.00019405767019586284,
      "loss": 1.1577,
      "step": 1901
    },
    {
      "epoch": 0.13710084336480935,
      "grad_norm": 3.219639755636662,
      "learning_rate": 0.00019404973937622727,
      "loss": 1.1353,
      "step": 1902
    },
    {
      "epoch": 0.13717292582714624,
      "grad_norm": 0.3035515436236788,
      "learning_rate": 0.00019404180343003953,
      "loss": 1.0686,
      "step": 1903
    },
    {
      "epoch": 0.13724500828948316,
      "grad_norm": 0.38517021122500283,
      "learning_rate": 0.00019403386235773216,
      "loss": 1.1257,
      "step": 1904
    },
    {
      "epoch": 0.13731709075182008,
      "grad_norm": 1.2874598183682677,
      "learning_rate": 0.000194025916159738,
      "loss": 1.1114,
      "step": 1905
    },
    {
      "epoch": 0.137389173214157,
      "grad_norm": 0.18217573309208132,
      "learning_rate": 0.0001940179648364903,
      "loss": 1.0626,
      "step": 1906
    },
    {
      "epoch": 0.13746125567649392,
      "grad_norm": 0.21430153240646416,
      "learning_rate": 0.00019401000838842238,
      "loss": 1.1191,
      "step": 1907
    },
    {
      "epoch": 0.1375333381388308,
      "grad_norm": 0.2076517361120412,
      "learning_rate": 0.00019400204681596796,
      "loss": 1.1066,
      "step": 1908
    },
    {
      "epoch": 0.13760542060116773,
      "grad_norm": 0.24349548542315005,
      "learning_rate": 0.00019399408011956098,
      "loss": 1.1645,
      "step": 1909
    },
    {
      "epoch": 0.13767750306350465,
      "grad_norm": 2.1692163205655697,
      "learning_rate": 0.00019398610829963583,
      "loss": 1.1391,
      "step": 1910
    },
    {
      "epoch": 0.13774958552584157,
      "grad_norm": 0.18784678482903056,
      "learning_rate": 0.00019397813135662692,
      "loss": 1.1436,
      "step": 1911
    },
    {
      "epoch": 0.13782166798817846,
      "grad_norm": 0.21102676423041763,
      "learning_rate": 0.00019397014929096903,
      "loss": 1.1138,
      "step": 1912
    },
    {
      "epoch": 0.13789375045051538,
      "grad_norm": 2.225885047760466,
      "learning_rate": 0.0001939621621030974,
      "loss": 1.1472,
      "step": 1913
    },
    {
      "epoch": 0.1379658329128523,
      "grad_norm": 2.72837065368223,
      "learning_rate": 0.00019395416979344732,
      "loss": 1.0938,
      "step": 1914
    },
    {
      "epoch": 0.13803791537518922,
      "grad_norm": 2.1443808033144363,
      "learning_rate": 0.0001939461723624544,
      "loss": 1.1785,
      "step": 1915
    },
    {
      "epoch": 0.13810999783752612,
      "grad_norm": 0.11786103881253805,
      "learning_rate": 0.00019393816981055467,
      "loss": 0.9968,
      "step": 1916
    },
    {
      "epoch": 0.13818208029986304,
      "grad_norm": 0.24538362428148314,
      "learning_rate": 0.00019393016213818423,
      "loss": 1.115,
      "step": 1917
    },
    {
      "epoch": 0.13825416276219996,
      "grad_norm": 0.37361921240633306,
      "learning_rate": 0.0001939221493457797,
      "loss": 1.0742,
      "step": 1918
    },
    {
      "epoch": 0.13832624522453688,
      "grad_norm": 0.357714292543503,
      "learning_rate": 0.00019391413143377773,
      "loss": 1.108,
      "step": 1919
    },
    {
      "epoch": 0.1383983276868738,
      "grad_norm": 0.11728404433545529,
      "learning_rate": 0.00019390610840261548,
      "loss": 1.1407,
      "step": 1920
    },
    {
      "epoch": 0.1384704101492107,
      "grad_norm": 0.15296461807737102,
      "learning_rate": 0.00019389808025273015,
      "loss": 1.1847,
      "step": 1921
    },
    {
      "epoch": 0.1385424926115476,
      "grad_norm": 0.13785349369886285,
      "learning_rate": 0.00019389004698455945,
      "loss": 1.1141,
      "step": 1922
    },
    {
      "epoch": 0.13861457507388453,
      "grad_norm": 0.12632259575269095,
      "learning_rate": 0.0001938820085985412,
      "loss": 1.0385,
      "step": 1923
    },
    {
      "epoch": 0.13868665753622145,
      "grad_norm": 2.2714521471405815,
      "learning_rate": 0.00019387396509511359,
      "loss": 1.1005,
      "step": 1924
    },
    {
      "epoch": 0.13875873999855834,
      "grad_norm": 0.16848843963752502,
      "learning_rate": 0.00019386591647471506,
      "loss": 1.0463,
      "step": 1925
    },
    {
      "epoch": 0.13883082246089526,
      "grad_norm": 5.071623487120747,
      "learning_rate": 0.00019385786273778434,
      "loss": 1.1523,
      "step": 1926
    },
    {
      "epoch": 0.13890290492323218,
      "grad_norm": 0.35980723507112417,
      "learning_rate": 0.00019384980388476044,
      "loss": 1.1105,
      "step": 1927
    },
    {
      "epoch": 0.1389749873855691,
      "grad_norm": 0.5579417230536807,
      "learning_rate": 0.0001938417399160826,
      "loss": 1.0838,
      "step": 1928
    },
    {
      "epoch": 0.139047069847906,
      "grad_norm": 0.15467289272461537,
      "learning_rate": 0.0001938336708321904,
      "loss": 1.0695,
      "step": 1929
    },
    {
      "epoch": 0.1391191523102429,
      "grad_norm": 1.7996789714431936,
      "learning_rate": 0.0001938255966335237,
      "loss": 1.1175,
      "step": 1930
    },
    {
      "epoch": 0.13919123477257983,
      "grad_norm": 0.2933148456592617,
      "learning_rate": 0.0001938175173205226,
      "loss": 1.1055,
      "step": 1931
    },
    {
      "epoch": 0.13926331723491675,
      "grad_norm": 0.252261405614959,
      "learning_rate": 0.00019380943289362747,
      "loss": 1.1795,
      "step": 1932
    },
    {
      "epoch": 0.13933539969725367,
      "grad_norm": 1.4921060334454144,
      "learning_rate": 0.000193801343353279,
      "loss": 1.1013,
      "step": 1933
    },
    {
      "epoch": 0.13940748215959056,
      "grad_norm": 0.3134692516071924,
      "learning_rate": 0.00019379324869991816,
      "loss": 1.0712,
      "step": 1934
    },
    {
      "epoch": 0.13947956462192748,
      "grad_norm": 0.14959083057382075,
      "learning_rate": 0.00019378514893398616,
      "loss": 1.1411,
      "step": 1935
    },
    {
      "epoch": 0.1395516470842644,
      "grad_norm": 0.21271474998545403,
      "learning_rate": 0.00019377704405592453,
      "loss": 1.0697,
      "step": 1936
    },
    {
      "epoch": 0.13962372954660132,
      "grad_norm": 0.47197128698180163,
      "learning_rate": 0.00019376893406617503,
      "loss": 1.0868,
      "step": 1937
    },
    {
      "epoch": 0.1396958120089382,
      "grad_norm": 2.021014672825603,
      "learning_rate": 0.00019376081896517976,
      "loss": 1.0775,
      "step": 1938
    },
    {
      "epoch": 0.13976789447127513,
      "grad_norm": 0.19193300313180936,
      "learning_rate": 0.00019375269875338105,
      "loss": 1.1158,
      "step": 1939
    },
    {
      "epoch": 0.13983997693361205,
      "grad_norm": 1.9942663495699284,
      "learning_rate": 0.00019374457343122149,
      "loss": 1.1458,
      "step": 1940
    },
    {
      "epoch": 0.13991205939594897,
      "grad_norm": 0.16911604858810253,
      "learning_rate": 0.00019373644299914403,
      "loss": 1.1129,
      "step": 1941
    },
    {
      "epoch": 0.13998414185828587,
      "grad_norm": 1.808678033570674,
      "learning_rate": 0.00019372830745759186,
      "loss": 1.1425,
      "step": 1942
    },
    {
      "epoch": 0.14005622432062279,
      "grad_norm": 0.13741057081576372,
      "learning_rate": 0.00019372016680700837,
      "loss": 1.1175,
      "step": 1943
    },
    {
      "epoch": 0.1401283067829597,
      "grad_norm": 0.5209863518446162,
      "learning_rate": 0.0001937120210478374,
      "loss": 1.0122,
      "step": 1944
    },
    {
      "epoch": 0.14020038924529662,
      "grad_norm": 0.6561203275517412,
      "learning_rate": 0.0001937038701805229,
      "loss": 1.1086,
      "step": 1945
    },
    {
      "epoch": 0.14027247170763354,
      "grad_norm": 0.20294710365762733,
      "learning_rate": 0.00019369571420550913,
      "loss": 1.1262,
      "step": 1946
    },
    {
      "epoch": 0.14034455416997044,
      "grad_norm": 0.970938020850956,
      "learning_rate": 0.00019368755312324076,
      "loss": 1.2018,
      "step": 1947
    },
    {
      "epoch": 0.14041663663230736,
      "grad_norm": 0.403782376983352,
      "learning_rate": 0.00019367938693416257,
      "loss": 0.9995,
      "step": 1948
    },
    {
      "epoch": 0.14048871909464428,
      "grad_norm": 1.3117780775854184,
      "learning_rate": 0.0001936712156387197,
      "loss": 1.1365,
      "step": 1949
    },
    {
      "epoch": 0.1405608015569812,
      "grad_norm": 0.48606897258179566,
      "learning_rate": 0.0001936630392373576,
      "loss": 1.1062,
      "step": 1950
    },
    {
      "epoch": 0.1406328840193181,
      "grad_norm": 0.24890342918168695,
      "learning_rate": 0.0001936548577305219,
      "loss": 1.1753,
      "step": 1951
    },
    {
      "epoch": 0.140704966481655,
      "grad_norm": 0.16086319287721623,
      "learning_rate": 0.00019364667111865862,
      "loss": 1.0877,
      "step": 1952
    },
    {
      "epoch": 0.14077704894399193,
      "grad_norm": 1.1466364372494129,
      "learning_rate": 0.00019363847940221396,
      "loss": 1.1051,
      "step": 1953
    },
    {
      "epoch": 0.14084913140632885,
      "grad_norm": 0.13168121713975095,
      "learning_rate": 0.00019363028258163447,
      "loss": 1.0841,
      "step": 1954
    },
    {
      "epoch": 0.14092121386866574,
      "grad_norm": 0.34350015520734095,
      "learning_rate": 0.00019362208065736695,
      "loss": 1.0712,
      "step": 1955
    },
    {
      "epoch": 0.14099329633100266,
      "grad_norm": 0.2984329327438169,
      "learning_rate": 0.0001936138736298584,
      "loss": 1.2129,
      "step": 1956
    },
    {
      "epoch": 0.14106537879333958,
      "grad_norm": 0.1817894234449243,
      "learning_rate": 0.00019360566149955632,
      "loss": 1.1514,
      "step": 1957
    },
    {
      "epoch": 0.1411374612556765,
      "grad_norm": 0.22094266537592505,
      "learning_rate": 0.00019359744426690822,
      "loss": 1.103,
      "step": 1958
    },
    {
      "epoch": 0.14120954371801342,
      "grad_norm": 0.1836238740509,
      "learning_rate": 0.00019358922193236212,
      "loss": 1.1052,
      "step": 1959
    },
    {
      "epoch": 0.1412816261803503,
      "grad_norm": 0.27846770668171317,
      "learning_rate": 0.00019358099449636612,
      "loss": 1.1114,
      "step": 1960
    },
    {
      "epoch": 0.14135370864268723,
      "grad_norm": 0.20497903654426597,
      "learning_rate": 0.00019357276195936868,
      "loss": 1.2039,
      "step": 1961
    },
    {
      "epoch": 0.14142579110502415,
      "grad_norm": 2.6069567723636564,
      "learning_rate": 0.00019356452432181862,
      "loss": 1.061,
      "step": 1962
    },
    {
      "epoch": 0.14149787356736107,
      "grad_norm": 0.5792951702896143,
      "learning_rate": 0.00019355628158416495,
      "loss": 1.1099,
      "step": 1963
    },
    {
      "epoch": 0.14156995602969796,
      "grad_norm": 0.795079634031427,
      "learning_rate": 0.00019354803374685694,
      "loss": 1.1517,
      "step": 1964
    },
    {
      "epoch": 0.14164203849203488,
      "grad_norm": 0.23974550874731415,
      "learning_rate": 0.00019353978081034417,
      "loss": 1.1535,
      "step": 1965
    },
    {
      "epoch": 0.1417141209543718,
      "grad_norm": 0.2772603889328082,
      "learning_rate": 0.00019353152277507652,
      "loss": 1.1295,
      "step": 1966
    },
    {
      "epoch": 0.14178620341670872,
      "grad_norm": 0.21021854473300533,
      "learning_rate": 0.00019352325964150415,
      "loss": 1.11,
      "step": 1967
    },
    {
      "epoch": 0.14185828587904561,
      "grad_norm": 0.1762179939723312,
      "learning_rate": 0.00019351499141007742,
      "loss": 1.2004,
      "step": 1968
    },
    {
      "epoch": 0.14193036834138253,
      "grad_norm": 0.12796694909975062,
      "learning_rate": 0.00019350671808124705,
      "loss": 1.1659,
      "step": 1969
    },
    {
      "epoch": 0.14200245080371945,
      "grad_norm": 0.09880947381159909,
      "learning_rate": 0.00019349843965546404,
      "loss": 1.1584,
      "step": 1970
    },
    {
      "epoch": 0.14207453326605637,
      "grad_norm": 2.775094909518855,
      "learning_rate": 0.00019349015613317957,
      "loss": 1.0712,
      "step": 1971
    },
    {
      "epoch": 0.1421466157283933,
      "grad_norm": 0.1380978378141614,
      "learning_rate": 0.00019348186751484524,
      "loss": 1.1132,
      "step": 1972
    },
    {
      "epoch": 0.1422186981907302,
      "grad_norm": 0.6544176389533791,
      "learning_rate": 0.00019347357380091279,
      "loss": 1.0987,
      "step": 1973
    },
    {
      "epoch": 0.1422907806530671,
      "grad_norm": 1.469258889379576,
      "learning_rate": 0.00019346527499183432,
      "loss": 1.181,
      "step": 1974
    },
    {
      "epoch": 0.14236286311540403,
      "grad_norm": 0.17034928089841384,
      "learning_rate": 0.00019345697108806223,
      "loss": 1.1696,
      "step": 1975
    },
    {
      "epoch": 0.14243494557774095,
      "grad_norm": 0.2032303018491949,
      "learning_rate": 0.00019344866209004908,
      "loss": 1.1154,
      "step": 1976
    },
    {
      "epoch": 0.14250702804007784,
      "grad_norm": 0.14907222488713437,
      "learning_rate": 0.00019344034799824789,
      "loss": 1.1593,
      "step": 1977
    },
    {
      "epoch": 0.14257911050241476,
      "grad_norm": 0.13620261386428534,
      "learning_rate": 0.00019343202881311173,
      "loss": 1.0995,
      "step": 1978
    },
    {
      "epoch": 0.14265119296475168,
      "grad_norm": 0.15608305563883915,
      "learning_rate": 0.00019342370453509419,
      "loss": 1.1083,
      "step": 1979
    },
    {
      "epoch": 0.1427232754270886,
      "grad_norm": 0.22598841886852825,
      "learning_rate": 0.00019341537516464894,
      "loss": 1.1116,
      "step": 1980
    },
    {
      "epoch": 0.1427953578894255,
      "grad_norm": 0.1184866329124749,
      "learning_rate": 0.00019340704070223,
      "loss": 1.1631,
      "step": 1981
    },
    {
      "epoch": 0.1428674403517624,
      "grad_norm": 0.11827908190157477,
      "learning_rate": 0.00019339870114829175,
      "loss": 1.0686,
      "step": 1982
    },
    {
      "epoch": 0.14293952281409933,
      "grad_norm": 2.0121777115288513,
      "learning_rate": 0.00019339035650328869,
      "loss": 1.2218,
      "step": 1983
    },
    {
      "epoch": 0.14301160527643625,
      "grad_norm": 0.3945061554615225,
      "learning_rate": 0.0001933820067676757,
      "loss": 1.0734,
      "step": 1984
    },
    {
      "epoch": 0.14308368773877317,
      "grad_norm": 0.14734086904022564,
      "learning_rate": 0.0001933736519419079,
      "loss": 1.065,
      "step": 1985
    },
    {
      "epoch": 0.14315577020111006,
      "grad_norm": 0.2955112420707221,
      "learning_rate": 0.0001933652920264408,
      "loss": 1.1678,
      "step": 1986
    },
    {
      "epoch": 0.14322785266344698,
      "grad_norm": 0.109816637521238,
      "learning_rate": 0.00019335692702172994,
      "loss": 1.1172,
      "step": 1987
    },
    {
      "epoch": 0.1432999351257839,
      "grad_norm": 0.16522432288811198,
      "learning_rate": 0.0001933485569282314,
      "loss": 1.0621,
      "step": 1988
    },
    {
      "epoch": 0.14337201758812082,
      "grad_norm": 0.13232375924263168,
      "learning_rate": 0.0001933401817464014,
      "loss": 1.1384,
      "step": 1989
    },
    {
      "epoch": 0.1434441000504577,
      "grad_norm": 0.47600020157584383,
      "learning_rate": 0.0001933318014766964,
      "loss": 1.17,
      "step": 1990
    },
    {
      "epoch": 0.14351618251279463,
      "grad_norm": 0.1324830400299979,
      "learning_rate": 0.00019332341611957325,
      "loss": 1.1407,
      "step": 1991
    },
    {
      "epoch": 0.14358826497513155,
      "grad_norm": 0.22165013116165438,
      "learning_rate": 0.00019331502567548908,
      "loss": 1.0685,
      "step": 1992
    },
    {
      "epoch": 0.14366034743746847,
      "grad_norm": 0.15114126465050584,
      "learning_rate": 0.00019330663014490115,
      "loss": 1.0825,
      "step": 1993
    },
    {
      "epoch": 0.14373242989980536,
      "grad_norm": 0.2527559307420724,
      "learning_rate": 0.00019329822952826714,
      "loss": 1.0911,
      "step": 1994
    },
    {
      "epoch": 0.14380451236214228,
      "grad_norm": 0.4954207517374255,
      "learning_rate": 0.00019328982382604498,
      "loss": 1.1208,
      "step": 1995
    },
    {
      "epoch": 0.1438765948244792,
      "grad_norm": 0.10236675671949202,
      "learning_rate": 0.0001932814130386928,
      "loss": 1.0718,
      "step": 1996
    },
    {
      "epoch": 0.14394867728681612,
      "grad_norm": 0.15633381949650352,
      "learning_rate": 0.0001932729971666691,
      "loss": 1.1083,
      "step": 1997
    },
    {
      "epoch": 0.14402075974915304,
      "grad_norm": 0.154869888206896,
      "learning_rate": 0.0001932645762104326,
      "loss": 1.1417,
      "step": 1998
    },
    {
      "epoch": 0.14409284221148994,
      "grad_norm": 0.14133052290141604,
      "learning_rate": 0.00019325615017044232,
      "loss": 1.2381,
      "step": 1999
    },
    {
      "epoch": 0.14416492467382686,
      "grad_norm": 0.1809273867853596,
      "learning_rate": 0.00019324771904715758,
      "loss": 1.1102,
      "step": 2000
    },
    {
      "epoch": 0.14423700713616378,
      "grad_norm": 0.11692041914134965,
      "learning_rate": 0.00019323928284103791,
      "loss": 1.0383,
      "step": 2001
    },
    {
      "epoch": 0.1443090895985007,
      "grad_norm": 0.14642557382502255,
      "learning_rate": 0.00019323084155254322,
      "loss": 1.018,
      "step": 2002
    },
    {
      "epoch": 0.1443811720608376,
      "grad_norm": 0.16760789885134914,
      "learning_rate": 0.00019322239518213357,
      "loss": 1.0638,
      "step": 2003
    },
    {
      "epoch": 0.1444532545231745,
      "grad_norm": 0.1964173447653102,
      "learning_rate": 0.0001932139437302694,
      "loss": 1.2031,
      "step": 2004
    },
    {
      "epoch": 0.14452533698551143,
      "grad_norm": 0.10295692065110147,
      "learning_rate": 0.00019320548719741138,
      "loss": 0.9972,
      "step": 2005
    },
    {
      "epoch": 0.14459741944784835,
      "grad_norm": 0.1266123143758802,
      "learning_rate": 0.00019319702558402046,
      "loss": 1.0767,
      "step": 2006
    },
    {
      "epoch": 0.14466950191018524,
      "grad_norm": 0.11867195369607703,
      "learning_rate": 0.0001931885588905579,
      "loss": 1.1087,
      "step": 2007
    },
    {
      "epoch": 0.14474158437252216,
      "grad_norm": 0.16942459683086694,
      "learning_rate": 0.0001931800871174852,
      "loss": 1.0945,
      "step": 2008
    },
    {
      "epoch": 0.14481366683485908,
      "grad_norm": 0.11995639696922986,
      "learning_rate": 0.0001931716102652641,
      "loss": 1.0935,
      "step": 2009
    },
    {
      "epoch": 0.144885749297196,
      "grad_norm": 0.10307456086547045,
      "learning_rate": 0.00019316312833435673,
      "loss": 1.098,
      "step": 2010
    },
    {
      "epoch": 0.14495783175953292,
      "grad_norm": 0.1279123327505083,
      "learning_rate": 0.0001931546413252254,
      "loss": 1.122,
      "step": 2011
    },
    {
      "epoch": 0.1450299142218698,
      "grad_norm": 0.14300896635101867,
      "learning_rate": 0.0001931461492383327,
      "loss": 1.0976,
      "step": 2012
    },
    {
      "epoch": 0.14510199668420673,
      "grad_norm": 0.14632480554519176,
      "learning_rate": 0.0001931376520741416,
      "loss": 1.0727,
      "step": 2013
    },
    {
      "epoch": 0.14517407914654365,
      "grad_norm": 0.10852848400166225,
      "learning_rate": 0.00019312914983311523,
      "loss": 1.048,
      "step": 2014
    },
    {
      "epoch": 0.14524616160888057,
      "grad_norm": 0.12215310825776471,
      "learning_rate": 0.00019312064251571703,
      "loss": 1.0973,
      "step": 2015
    },
    {
      "epoch": 0.14531824407121746,
      "grad_norm": 0.14127858163517637,
      "learning_rate": 0.0001931121301224107,
      "loss": 1.1231,
      "step": 2016
    },
    {
      "epoch": 0.14539032653355438,
      "grad_norm": 0.10271772133273015,
      "learning_rate": 0.00019310361265366033,
      "loss": 1.0506,
      "step": 2017
    },
    {
      "epoch": 0.1454624089958913,
      "grad_norm": 0.13681590022889373,
      "learning_rate": 0.0001930950901099301,
      "loss": 1.1071,
      "step": 2018
    },
    {
      "epoch": 0.14553449145822822,
      "grad_norm": 0.10790260216964392,
      "learning_rate": 0.0001930865624916846,
      "loss": 1.0671,
      "step": 2019
    },
    {
      "epoch": 0.1456065739205651,
      "grad_norm": 2.5020728239784824,
      "learning_rate": 0.0001930780297993887,
      "loss": 1.0688,
      "step": 2020
    },
    {
      "epoch": 0.14567865638290203,
      "grad_norm": 0.15375119553833502,
      "learning_rate": 0.00019306949203350747,
      "loss": 1.1506,
      "step": 2021
    },
    {
      "epoch": 0.14575073884523895,
      "grad_norm": 0.12214537203139474,
      "learning_rate": 0.0001930609491945063,
      "loss": 1.0937,
      "step": 2022
    },
    {
      "epoch": 0.14582282130757587,
      "grad_norm": 0.10435458822330212,
      "learning_rate": 0.00019305240128285081,
      "loss": 1.1265,
      "step": 2023
    },
    {
      "epoch": 0.1458949037699128,
      "grad_norm": 0.17097755240711462,
      "learning_rate": 0.00019304384829900702,
      "loss": 1.055,
      "step": 2024
    },
    {
      "epoch": 0.14596698623224968,
      "grad_norm": 0.6068272277810789,
      "learning_rate": 0.0001930352902434411,
      "loss": 1.1104,
      "step": 2025
    },
    {
      "epoch": 0.1460390686945866,
      "grad_norm": 0.16793640265913112,
      "learning_rate": 0.00019302672711661955,
      "loss": 1.0567,
      "step": 2026
    },
    {
      "epoch": 0.14611115115692352,
      "grad_norm": 0.1439407392151587,
      "learning_rate": 0.00019301815891900912,
      "loss": 1.0562,
      "step": 2027
    },
    {
      "epoch": 0.14618323361926044,
      "grad_norm": 0.1659698147751649,
      "learning_rate": 0.00019300958565107688,
      "loss": 1.1094,
      "step": 2028
    },
    {
      "epoch": 0.14625531608159734,
      "grad_norm": 0.16607531083022578,
      "learning_rate": 0.00019300100731329013,
      "loss": 1.0979,
      "step": 2029
    },
    {
      "epoch": 0.14632739854393426,
      "grad_norm": 0.15562619062244343,
      "learning_rate": 0.00019299242390611644,
      "loss": 1.0445,
      "step": 2030
    },
    {
      "epoch": 0.14639948100627118,
      "grad_norm": 0.1241505885149626,
      "learning_rate": 0.00019298383543002375,
      "loss": 1.0933,
      "step": 2031
    },
    {
      "epoch": 0.1464715634686081,
      "grad_norm": 0.15576446467887173,
      "learning_rate": 0.00019297524188548017,
      "loss": 1.1426,
      "step": 2032
    },
    {
      "epoch": 0.146543645930945,
      "grad_norm": 0.1543074655938027,
      "learning_rate": 0.00019296664327295412,
      "loss": 1.1409,
      "step": 2033
    },
    {
      "epoch": 0.1466157283932819,
      "grad_norm": 0.12950897076215923,
      "learning_rate": 0.0001929580395929143,
      "loss": 1.0636,
      "step": 2034
    },
    {
      "epoch": 0.14668781085561883,
      "grad_norm": 0.1282161254393729,
      "learning_rate": 0.00019294943084582967,
      "loss": 1.0737,
      "step": 2035
    },
    {
      "epoch": 0.14675989331795575,
      "grad_norm": 0.19350413919445725,
      "learning_rate": 0.00019294081703216955,
      "loss": 1.1423,
      "step": 2036
    },
    {
      "epoch": 0.14683197578029267,
      "grad_norm": 0.106609817019522,
      "learning_rate": 0.00019293219815240344,
      "loss": 1.1628,
      "step": 2037
    },
    {
      "epoch": 0.14690405824262956,
      "grad_norm": 0.11921472109339457,
      "learning_rate": 0.0001929235742070011,
      "loss": 1.0197,
      "step": 2038
    },
    {
      "epoch": 0.14697614070496648,
      "grad_norm": 0.16862880874543396,
      "learning_rate": 0.00019291494519643268,
      "loss": 1.1774,
      "step": 2039
    },
    {
      "epoch": 0.1470482231673034,
      "grad_norm": 0.1307923662782066,
      "learning_rate": 0.00019290631112116845,
      "loss": 1.1753,
      "step": 2040
    },
    {
      "epoch": 0.14712030562964032,
      "grad_norm": 0.12037116995150901,
      "learning_rate": 0.00019289767198167916,
      "loss": 1.1016,
      "step": 2041
    },
    {
      "epoch": 0.1471923880919772,
      "grad_norm": 0.11679505741253579,
      "learning_rate": 0.00019288902777843563,
      "loss": 1.114,
      "step": 2042
    },
    {
      "epoch": 0.14726447055431413,
      "grad_norm": 0.12974797818304107,
      "learning_rate": 0.0001928803785119091,
      "loss": 1.1096,
      "step": 2043
    },
    {
      "epoch": 0.14733655301665105,
      "grad_norm": 0.12037650273730191,
      "learning_rate": 0.00019287172418257098,
      "loss": 1.0155,
      "step": 2044
    },
    {
      "epoch": 0.14740863547898797,
      "grad_norm": 0.12135984403716508,
      "learning_rate": 0.00019286306479089307,
      "loss": 1.125,
      "step": 2045
    },
    {
      "epoch": 0.14748071794132486,
      "grad_norm": 0.1186774502830935,
      "learning_rate": 0.00019285440033734732,
      "loss": 1.1973,
      "step": 2046
    },
    {
      "epoch": 0.14755280040366178,
      "grad_norm": 0.1492560680272679,
      "learning_rate": 0.00019284573082240607,
      "loss": 1.1525,
      "step": 2047
    },
    {
      "epoch": 0.1476248828659987,
      "grad_norm": 0.11375070641019172,
      "learning_rate": 0.00019283705624654186,
      "loss": 1.0896,
      "step": 2048
    },
    {
      "epoch": 0.14769696532833562,
      "grad_norm": 0.14124477340943886,
      "learning_rate": 0.00019282837661022755,
      "loss": 1.0504,
      "step": 2049
    },
    {
      "epoch": 0.14776904779067254,
      "grad_norm": 0.112480780381201,
      "learning_rate": 0.00019281969191393622,
      "loss": 1.0637,
      "step": 2050
    },
    {
      "epoch": 0.14784113025300943,
      "grad_norm": 0.19483928818088425,
      "learning_rate": 0.00019281100215814135,
      "loss": 1.112,
      "step": 2051
    },
    {
      "epoch": 0.14791321271534635,
      "grad_norm": 0.11727300256620093,
      "learning_rate": 0.0001928023073433165,
      "loss": 1.1599,
      "step": 2052
    },
    {
      "epoch": 0.14798529517768327,
      "grad_norm": 0.3435286893849854,
      "learning_rate": 0.00019279360746993568,
      "loss": 1.1449,
      "step": 2053
    },
    {
      "epoch": 0.1480573776400202,
      "grad_norm": 0.1157745817313348,
      "learning_rate": 0.0001927849025384731,
      "loss": 1.0858,
      "step": 2054
    },
    {
      "epoch": 0.14812946010235709,
      "grad_norm": 0.12200315098166976,
      "learning_rate": 0.00019277619254940326,
      "loss": 1.1121,
      "step": 2055
    },
    {
      "epoch": 0.148201542564694,
      "grad_norm": 0.11374293269363607,
      "learning_rate": 0.00019276747750320095,
      "loss": 1.0811,
      "step": 2056
    },
    {
      "epoch": 0.14827362502703093,
      "grad_norm": 0.18774768334147798,
      "learning_rate": 0.00019275875740034116,
      "loss": 1.1108,
      "step": 2057
    },
    {
      "epoch": 0.14834570748936785,
      "grad_norm": 0.1473830539564324,
      "learning_rate": 0.00019275003224129925,
      "loss": 1.0865,
      "step": 2058
    },
    {
      "epoch": 0.14841778995170474,
      "grad_norm": 1.7195594607353437,
      "learning_rate": 0.0001927413020265508,
      "loss": 1.1088,
      "step": 2059
    },
    {
      "epoch": 0.14848987241404166,
      "grad_norm": 0.19063276236301296,
      "learning_rate": 0.00019273256675657175,
      "loss": 1.1274,
      "step": 2060
    },
    {
      "epoch": 0.14856195487637858,
      "grad_norm": 0.17142589645429868,
      "learning_rate": 0.00019272382643183817,
      "loss": 1.1527,
      "step": 2061
    },
    {
      "epoch": 0.1486340373387155,
      "grad_norm": 0.12512884790313333,
      "learning_rate": 0.00019271508105282652,
      "loss": 1.1226,
      "step": 2062
    },
    {
      "epoch": 0.14870611980105242,
      "grad_norm": 0.10494694483758155,
      "learning_rate": 0.0001927063306200135,
      "loss": 1.0057,
      "step": 2063
    },
    {
      "epoch": 0.1487782022633893,
      "grad_norm": 0.1010855428451534,
      "learning_rate": 0.00019269757513387613,
      "loss": 1.1128,
      "step": 2064
    },
    {
      "epoch": 0.14885028472572623,
      "grad_norm": 0.16188459661400725,
      "learning_rate": 0.00019268881459489156,
      "loss": 1.0748,
      "step": 2065
    },
    {
      "epoch": 0.14892236718806315,
      "grad_norm": 0.12811627869510575,
      "learning_rate": 0.0001926800490035374,
      "loss": 1.1038,
      "step": 2066
    },
    {
      "epoch": 0.14899444965040007,
      "grad_norm": 0.09635309590471935,
      "learning_rate": 0.0001926712783602914,
      "loss": 1.039,
      "step": 2067
    },
    {
      "epoch": 0.14906653211273696,
      "grad_norm": 0.4334394878916393,
      "learning_rate": 0.00019266250266563174,
      "loss": 1.0982,
      "step": 2068
    },
    {
      "epoch": 0.14913861457507388,
      "grad_norm": 0.10262923739176551,
      "learning_rate": 0.00019265372192003665,
      "loss": 1.0806,
      "step": 2069
    },
    {
      "epoch": 0.1492106970374108,
      "grad_norm": 0.1273087983180443,
      "learning_rate": 0.00019264493612398481,
      "loss": 1.2109,
      "step": 2070
    },
    {
      "epoch": 0.14928277949974772,
      "grad_norm": 0.22618191055173625,
      "learning_rate": 0.00019263614527795513,
      "loss": 1.0872,
      "step": 2071
    },
    {
      "epoch": 0.1493548619620846,
      "grad_norm": 0.1573671926071853,
      "learning_rate": 0.00019262734938242678,
      "loss": 1.106,
      "step": 2072
    },
    {
      "epoch": 0.14942694442442153,
      "grad_norm": 0.416230355908708,
      "learning_rate": 0.00019261854843787926,
      "loss": 1.0175,
      "step": 2073
    },
    {
      "epoch": 0.14949902688675845,
      "grad_norm": 0.15308110134507613,
      "learning_rate": 0.00019260974244479222,
      "loss": 1.0995,
      "step": 2074
    },
    {
      "epoch": 0.14957110934909537,
      "grad_norm": 1.5665077281358946,
      "learning_rate": 0.00019260093140364572,
      "loss": 1.2245,
      "step": 2075
    },
    {
      "epoch": 0.1496431918114323,
      "grad_norm": 1.1124568326979942,
      "learning_rate": 0.00019259211531492002,
      "loss": 1.066,
      "step": 2076
    },
    {
      "epoch": 0.14971527427376918,
      "grad_norm": 0.1244254237484156,
      "learning_rate": 0.0001925832941790957,
      "loss": 1.0947,
      "step": 2077
    },
    {
      "epoch": 0.1497873567361061,
      "grad_norm": 0.2943340821327353,
      "learning_rate": 0.00019257446799665355,
      "loss": 1.165,
      "step": 2078
    },
    {
      "epoch": 0.14985943919844302,
      "grad_norm": 0.13565017577699695,
      "learning_rate": 0.00019256563676807472,
      "loss": 1.0616,
      "step": 2079
    },
    {
      "epoch": 0.14993152166077994,
      "grad_norm": 0.12245256737386284,
      "learning_rate": 0.00019255680049384054,
      "loss": 1.1358,
      "step": 2080
    },
    {
      "epoch": 0.15000360412311683,
      "grad_norm": 0.1184549117411675,
      "learning_rate": 0.00019254795917443275,
      "loss": 1.0636,
      "step": 2081
    },
    {
      "epoch": 0.15007568658545375,
      "grad_norm": 0.7352552355105769,
      "learning_rate": 0.00019253911281033319,
      "loss": 1.1534,
      "step": 2082
    },
    {
      "epoch": 0.15014776904779067,
      "grad_norm": 0.16394601875327894,
      "learning_rate": 0.0001925302614020241,
      "loss": 1.1414,
      "step": 2083
    },
    {
      "epoch": 0.1502198515101276,
      "grad_norm": 0.1260383263609447,
      "learning_rate": 0.00019252140494998798,
      "loss": 1.0817,
      "step": 2084
    },
    {
      "epoch": 0.1502919339724645,
      "grad_norm": 0.12792021788905808,
      "learning_rate": 0.00019251254345470758,
      "loss": 1.0556,
      "step": 2085
    },
    {
      "epoch": 0.1503640164348014,
      "grad_norm": 0.228345268417868,
      "learning_rate": 0.00019250367691666588,
      "loss": 1.1333,
      "step": 2086
    },
    {
      "epoch": 0.15043609889713833,
      "grad_norm": 0.11268774039410336,
      "learning_rate": 0.00019249480533634628,
      "loss": 1.108,
      "step": 2087
    },
    {
      "epoch": 0.15050818135947525,
      "grad_norm": 0.12180125668114185,
      "learning_rate": 0.00019248592871423225,
      "loss": 1.0456,
      "step": 2088
    },
    {
      "epoch": 0.15058026382181217,
      "grad_norm": 0.11721546523945515,
      "learning_rate": 0.00019247704705080773,
      "loss": 1.1354,
      "step": 2089
    },
    {
      "epoch": 0.15065234628414906,
      "grad_norm": 0.1448070027034231,
      "learning_rate": 0.00019246816034655683,
      "loss": 1.0411,
      "step": 2090
    },
    {
      "epoch": 0.15072442874648598,
      "grad_norm": 0.15381577860245146,
      "learning_rate": 0.00019245926860196396,
      "loss": 1.1319,
      "step": 2091
    },
    {
      "epoch": 0.1507965112088229,
      "grad_norm": 0.22999081704771318,
      "learning_rate": 0.00019245037181751375,
      "loss": 0.9997,
      "step": 2092
    },
    {
      "epoch": 0.15086859367115982,
      "grad_norm": 0.12246705931299459,
      "learning_rate": 0.00019244146999369122,
      "loss": 1.1029,
      "step": 2093
    },
    {
      "epoch": 0.1509406761334967,
      "grad_norm": 0.12574491421860046,
      "learning_rate": 0.00019243256313098155,
      "loss": 1.1493,
      "step": 2094
    },
    {
      "epoch": 0.15101275859583363,
      "grad_norm": 0.13873866147787603,
      "learning_rate": 0.00019242365122987027,
      "loss": 1.0921,
      "step": 2095
    },
    {
      "epoch": 0.15108484105817055,
      "grad_norm": 0.10381449987877078,
      "learning_rate": 0.00019241473429084315,
      "loss": 1.0504,
      "step": 2096
    },
    {
      "epoch": 0.15115692352050747,
      "grad_norm": 0.14214742623312526,
      "learning_rate": 0.00019240581231438626,
      "loss": 1.2301,
      "step": 2097
    },
    {
      "epoch": 0.15122900598284436,
      "grad_norm": 0.11478963082984019,
      "learning_rate": 0.0001923968853009859,
      "loss": 1.0943,
      "step": 2098
    },
    {
      "epoch": 0.15130108844518128,
      "grad_norm": 0.11090356174489908,
      "learning_rate": 0.0001923879532511287,
      "loss": 1.0773,
      "step": 2099
    },
    {
      "epoch": 0.1513731709075182,
      "grad_norm": 0.28288183118720245,
      "learning_rate": 0.0001923790161653015,
      "loss": 1.0889,
      "step": 2100
    },
    {
      "epoch": 0.15144525336985512,
      "grad_norm": 0.1276372342271995,
      "learning_rate": 0.00019237007404399148,
      "loss": 1.0793,
      "step": 2101
    },
    {
      "epoch": 0.15151733583219204,
      "grad_norm": 0.2598184530597864,
      "learning_rate": 0.00019236112688768606,
      "loss": 1.1048,
      "step": 2102
    },
    {
      "epoch": 0.15158941829452893,
      "grad_norm": 0.12381829105626486,
      "learning_rate": 0.00019235217469687293,
      "loss": 1.0425,
      "step": 2103
    },
    {
      "epoch": 0.15166150075686585,
      "grad_norm": 0.1165880782516636,
      "learning_rate": 0.0001923432174720401,
      "loss": 1.1175,
      "step": 2104
    },
    {
      "epoch": 0.15173358321920277,
      "grad_norm": 0.13192536120426793,
      "learning_rate": 0.00019233425521367577,
      "loss": 1.1085,
      "step": 2105
    },
    {
      "epoch": 0.1518056656815397,
      "grad_norm": 0.11774863300163792,
      "learning_rate": 0.00019232528792226846,
      "loss": 1.069,
      "step": 2106
    },
    {
      "epoch": 0.15187774814387658,
      "grad_norm": 0.1531316542092164,
      "learning_rate": 0.00019231631559830703,
      "loss": 1.0979,
      "step": 2107
    },
    {
      "epoch": 0.1519498306062135,
      "grad_norm": 0.16343391815474512,
      "learning_rate": 0.0001923073382422805,
      "loss": 1.0845,
      "step": 2108
    },
    {
      "epoch": 0.15202191306855042,
      "grad_norm": 0.13497112979151565,
      "learning_rate": 0.00019229835585467825,
      "loss": 1.1304,
      "step": 2109
    },
    {
      "epoch": 0.15209399553088734,
      "grad_norm": 0.7168631291711581,
      "learning_rate": 0.0001922893684359899,
      "loss": 1.0953,
      "step": 2110
    },
    {
      "epoch": 0.15216607799322424,
      "grad_norm": 0.1477407804071476,
      "learning_rate": 0.00019228037598670525,
      "loss": 1.0391,
      "step": 2111
    },
    {
      "epoch": 0.15223816045556116,
      "grad_norm": 0.17013760165134104,
      "learning_rate": 0.0001922713785073146,
      "loss": 1.1569,
      "step": 2112
    },
    {
      "epoch": 0.15231024291789808,
      "grad_norm": 0.40488933591333354,
      "learning_rate": 0.00019226237599830834,
      "loss": 1.1579,
      "step": 2113
    },
    {
      "epoch": 0.152382325380235,
      "grad_norm": 0.1262762544754975,
      "learning_rate": 0.00019225336846017717,
      "loss": 1.0729,
      "step": 2114
    },
    {
      "epoch": 0.15245440784257192,
      "grad_norm": 0.12222885373778965,
      "learning_rate": 0.00019224435589341207,
      "loss": 1.174,
      "step": 2115
    },
    {
      "epoch": 0.1525264903049088,
      "grad_norm": 0.1326233404786694,
      "learning_rate": 0.0001922353382985044,
      "loss": 1.067,
      "step": 2116
    },
    {
      "epoch": 0.15259857276724573,
      "grad_norm": 0.12376891853239351,
      "learning_rate": 0.00019222631567594557,
      "loss": 1.1062,
      "step": 2117
    },
    {
      "epoch": 0.15267065522958265,
      "grad_norm": 0.12858916900873935,
      "learning_rate": 0.0001922172880262275,
      "loss": 1.077,
      "step": 2118
    },
    {
      "epoch": 0.15274273769191957,
      "grad_norm": 0.194884744966175,
      "learning_rate": 0.00019220825534984218,
      "loss": 1.1198,
      "step": 2119
    },
    {
      "epoch": 0.15281482015425646,
      "grad_norm": 0.15091982122771735,
      "learning_rate": 0.00019219921764728208,
      "loss": 0.9988,
      "step": 2120
    },
    {
      "epoch": 0.15288690261659338,
      "grad_norm": 0.4429217075570465,
      "learning_rate": 0.00019219017491903974,
      "loss": 1.1037,
      "step": 2121
    },
    {
      "epoch": 0.1529589850789303,
      "grad_norm": 0.1005172653570972,
      "learning_rate": 0.00019218112716560813,
      "loss": 1.1323,
      "step": 2122
    },
    {
      "epoch": 0.15303106754126722,
      "grad_norm": 0.12891666346780742,
      "learning_rate": 0.00019217207438748042,
      "loss": 1.1851,
      "step": 2123
    },
    {
      "epoch": 0.1531031500036041,
      "grad_norm": 0.15240635467775607,
      "learning_rate": 0.00019216301658515005,
      "loss": 1.1118,
      "step": 2124
    },
    {
      "epoch": 0.15317523246594103,
      "grad_norm": 1.8560548166805413,
      "learning_rate": 0.00019215395375911073,
      "loss": 1.0958,
      "step": 2125
    },
    {
      "epoch": 0.15324731492827795,
      "grad_norm": 0.1752306858842939,
      "learning_rate": 0.00019214488590985656,
      "loss": 1.1277,
      "step": 2126
    },
    {
      "epoch": 0.15331939739061487,
      "grad_norm": 0.10712285671261333,
      "learning_rate": 0.00019213581303788172,
      "loss": 1.1754,
      "step": 2127
    },
    {
      "epoch": 0.1533914798529518,
      "grad_norm": 0.12018421363991574,
      "learning_rate": 0.0001921267351436808,
      "loss": 1.1291,
      "step": 2128
    },
    {
      "epoch": 0.15346356231528868,
      "grad_norm": 0.13769129598814106,
      "learning_rate": 0.00019211765222774865,
      "loss": 1.0649,
      "step": 2129
    },
    {
      "epoch": 0.1535356447776256,
      "grad_norm": 0.12345693537861563,
      "learning_rate": 0.0001921085642905803,
      "loss": 1.1188,
      "step": 2130
    },
    {
      "epoch": 0.15360772723996252,
      "grad_norm": 0.14054635517288502,
      "learning_rate": 0.0001920994713326712,
      "loss": 1.1079,
      "step": 2131
    },
    {
      "epoch": 0.15367980970229944,
      "grad_norm": 1.0563226930489935,
      "learning_rate": 0.00019209037335451697,
      "loss": 0.9832,
      "step": 2132
    },
    {
      "epoch": 0.15375189216463633,
      "grad_norm": 1.3214432982231366,
      "learning_rate": 0.00019208127035661354,
      "loss": 1.127,
      "step": 2133
    },
    {
      "epoch": 0.15382397462697325,
      "grad_norm": 2.1718585162141366,
      "learning_rate": 0.00019207216233945706,
      "loss": 1.1365,
      "step": 2134
    },
    {
      "epoch": 0.15389605708931017,
      "grad_norm": 0.11878985142264462,
      "learning_rate": 0.00019206304930354404,
      "loss": 1.1493,
      "step": 2135
    },
    {
      "epoch": 0.1539681395516471,
      "grad_norm": 0.2439501589059495,
      "learning_rate": 0.0001920539312493712,
      "loss": 1.1667,
      "step": 2136
    },
    {
      "epoch": 0.15404022201398399,
      "grad_norm": 0.1687473467452523,
      "learning_rate": 0.0001920448081774356,
      "loss": 1.0883,
      "step": 2137
    },
    {
      "epoch": 0.1541123044763209,
      "grad_norm": 0.3870598058349409,
      "learning_rate": 0.0001920356800882345,
      "loss": 1.1941,
      "step": 2138
    },
    {
      "epoch": 0.15418438693865782,
      "grad_norm": 1.4124446382839768,
      "learning_rate": 0.00019202654698226542,
      "loss": 1.0933,
      "step": 2139
    },
    {
      "epoch": 0.15425646940099474,
      "grad_norm": 0.10354971942644708,
      "learning_rate": 0.0001920174088600263,
      "loss": 1.0032,
      "step": 2140
    },
    {
      "epoch": 0.15432855186333166,
      "grad_norm": 0.1672650239561917,
      "learning_rate": 0.00019200826572201513,
      "loss": 1.0736,
      "step": 2141
    },
    {
      "epoch": 0.15440063432566856,
      "grad_norm": 0.5317303553111726,
      "learning_rate": 0.00019199911756873038,
      "loss": 1.1522,
      "step": 2142
    },
    {
      "epoch": 0.15447271678800548,
      "grad_norm": 0.17851984994357106,
      "learning_rate": 0.00019198996440067064,
      "loss": 1.1114,
      "step": 2143
    },
    {
      "epoch": 0.1545447992503424,
      "grad_norm": 0.18538734042225158,
      "learning_rate": 0.0001919808062183349,
      "loss": 1.0946,
      "step": 2144
    },
    {
      "epoch": 0.15461688171267932,
      "grad_norm": 0.3890787377601641,
      "learning_rate": 0.00019197164302222234,
      "loss": 1.1066,
      "step": 2145
    },
    {
      "epoch": 0.1546889641750162,
      "grad_norm": 0.1312208759759674,
      "learning_rate": 0.00019196247481283242,
      "loss": 1.0275,
      "step": 2146
    },
    {
      "epoch": 0.15476104663735313,
      "grad_norm": 0.126502925985371,
      "learning_rate": 0.00019195330159066493,
      "loss": 1.0637,
      "step": 2147
    },
    {
      "epoch": 0.15483312909969005,
      "grad_norm": 0.1362837552600864,
      "learning_rate": 0.00019194412335621984,
      "loss": 1.082,
      "step": 2148
    },
    {
      "epoch": 0.15490521156202697,
      "grad_norm": 0.13566417540250722,
      "learning_rate": 0.00019193494010999747,
      "loss": 1.1464,
      "step": 2149
    },
    {
      "epoch": 0.15497729402436386,
      "grad_norm": 0.2779150335740342,
      "learning_rate": 0.00019192575185249838,
      "loss": 1.1098,
      "step": 2150
    },
    {
      "epoch": 0.15504937648670078,
      "grad_norm": 0.37446925783709684,
      "learning_rate": 0.00019191655858422346,
      "loss": 1.1309,
      "step": 2151
    },
    {
      "epoch": 0.1551214589490377,
      "grad_norm": 2.2666444899687037,
      "learning_rate": 0.00019190736030567376,
      "loss": 1.1557,
      "step": 2152
    },
    {
      "epoch": 0.15519354141137462,
      "grad_norm": 0.17945403555954131,
      "learning_rate": 0.00019189815701735073,
      "loss": 1.0748,
      "step": 2153
    },
    {
      "epoch": 0.15526562387371154,
      "grad_norm": 0.21910395714034922,
      "learning_rate": 0.00019188894871975594,
      "loss": 1.0838,
      "step": 2154
    },
    {
      "epoch": 0.15533770633604843,
      "grad_norm": 0.2356518829443706,
      "learning_rate": 0.00019187973541339145,
      "loss": 1.1042,
      "step": 2155
    },
    {
      "epoch": 0.15540978879838535,
      "grad_norm": 0.09247157729226117,
      "learning_rate": 0.00019187051709875936,
      "loss": 1.1424,
      "step": 2156
    },
    {
      "epoch": 0.15548187126072227,
      "grad_norm": 0.34266616484383466,
      "learning_rate": 0.0001918612937763622,
      "loss": 1.0662,
      "step": 2157
    },
    {
      "epoch": 0.1555539537230592,
      "grad_norm": 0.143884249984806,
      "learning_rate": 0.0001918520654467027,
      "loss": 1.0581,
      "step": 2158
    },
    {
      "epoch": 0.15562603618539608,
      "grad_norm": 0.1304632410915559,
      "learning_rate": 0.0001918428321102839,
      "loss": 1.0947,
      "step": 2159
    },
    {
      "epoch": 0.155698118647733,
      "grad_norm": 0.13449671552044817,
      "learning_rate": 0.00019183359376760911,
      "loss": 1.1482,
      "step": 2160
    },
    {
      "epoch": 0.15577020111006992,
      "grad_norm": 0.23732691419903398,
      "learning_rate": 0.0001918243504191819,
      "loss": 1.0847,
      "step": 2161
    },
    {
      "epoch": 0.15584228357240684,
      "grad_norm": 0.13016795989796592,
      "learning_rate": 0.00019181510206550613,
      "loss": 1.1104,
      "step": 2162
    },
    {
      "epoch": 0.15591436603474373,
      "grad_norm": 0.33323141054694594,
      "learning_rate": 0.00019180584870708586,
      "loss": 1.0623,
      "step": 2163
    },
    {
      "epoch": 0.15598644849708065,
      "grad_norm": 0.1119979290222621,
      "learning_rate": 0.00019179659034442552,
      "loss": 1.1369,
      "step": 2164
    },
    {
      "epoch": 0.15605853095941757,
      "grad_norm": 0.7682595338815409,
      "learning_rate": 0.0001917873269780298,
      "loss": 1.1489,
      "step": 2165
    },
    {
      "epoch": 0.1561306134217545,
      "grad_norm": 0.13023350687256388,
      "learning_rate": 0.00019177805860840356,
      "loss": 1.0625,
      "step": 2166
    },
    {
      "epoch": 0.1562026958840914,
      "grad_norm": 0.34622653149289434,
      "learning_rate": 0.00019176878523605208,
      "loss": 1.1586,
      "step": 2167
    },
    {
      "epoch": 0.1562747783464283,
      "grad_norm": 0.141868084524279,
      "learning_rate": 0.0001917595068614808,
      "loss": 1.0922,
      "step": 2168
    },
    {
      "epoch": 0.15634686080876523,
      "grad_norm": 0.3199307692690403,
      "learning_rate": 0.00019175022348519554,
      "loss": 1.0564,
      "step": 2169
    },
    {
      "epoch": 0.15641894327110215,
      "grad_norm": 0.2870724293569073,
      "learning_rate": 0.00019174093510770224,
      "loss": 1.1941,
      "step": 2170
    },
    {
      "epoch": 0.15649102573343907,
      "grad_norm": 0.11187465494145134,
      "learning_rate": 0.00019173164172950723,
      "loss": 1.139,
      "step": 2171
    },
    {
      "epoch": 0.15656310819577596,
      "grad_norm": 0.16499730976710344,
      "learning_rate": 0.00019172234335111714,
      "loss": 1.104,
      "step": 2172
    },
    {
      "epoch": 0.15663519065811288,
      "grad_norm": 0.10658344730528063,
      "learning_rate": 0.00019171303997303873,
      "loss": 1.1579,
      "step": 2173
    },
    {
      "epoch": 0.1567072731204498,
      "grad_norm": 0.11958493501634854,
      "learning_rate": 0.00019170373159577917,
      "loss": 1.06,
      "step": 2174
    },
    {
      "epoch": 0.15677935558278672,
      "grad_norm": 0.2222029372854195,
      "learning_rate": 0.0001916944182198458,
      "loss": 1.1337,
      "step": 2175
    },
    {
      "epoch": 0.1568514380451236,
      "grad_norm": 0.14021532384720237,
      "learning_rate": 0.00019168509984574634,
      "loss": 1.1263,
      "step": 2176
    },
    {
      "epoch": 0.15692352050746053,
      "grad_norm": 0.272829828785307,
      "learning_rate": 0.00019167577647398871,
      "loss": 1.0679,
      "step": 2177
    },
    {
      "epoch": 0.15699560296979745,
      "grad_norm": 1.3556048956271887,
      "learning_rate": 0.00019166644810508107,
      "loss": 1.1595,
      "step": 2178
    },
    {
      "epoch": 0.15706768543213437,
      "grad_norm": 0.22280393564997147,
      "learning_rate": 0.00019165711473953197,
      "loss": 1.1429,
      "step": 2179
    },
    {
      "epoch": 0.1571397678944713,
      "grad_norm": 0.15118064939687054,
      "learning_rate": 0.00019164777637785011,
      "loss": 1.0821,
      "step": 2180
    },
    {
      "epoch": 0.15721185035680818,
      "grad_norm": 0.16653940738439013,
      "learning_rate": 0.00019163843302054455,
      "loss": 1.0995,
      "step": 2181
    },
    {
      "epoch": 0.1572839328191451,
      "grad_norm": 0.26271469186327456,
      "learning_rate": 0.00019162908466812458,
      "loss": 1.1285,
      "step": 2182
    },
    {
      "epoch": 0.15735601528148202,
      "grad_norm": 0.16333898620360257,
      "learning_rate": 0.0001916197313210997,
      "loss": 1.0716,
      "step": 2183
    },
    {
      "epoch": 0.15742809774381894,
      "grad_norm": 0.1186830567484475,
      "learning_rate": 0.00019161037297997987,
      "loss": 1.1072,
      "step": 2184
    },
    {
      "epoch": 0.15750018020615583,
      "grad_norm": 0.13341905488659575,
      "learning_rate": 0.00019160100964527512,
      "loss": 1.0847,
      "step": 2185
    },
    {
      "epoch": 0.15757226266849275,
      "grad_norm": 0.1452190638759728,
      "learning_rate": 0.00019159164131749587,
      "loss": 1.1281,
      "step": 2186
    },
    {
      "epoch": 0.15764434513082967,
      "grad_norm": 0.11327130589704089,
      "learning_rate": 0.00019158226799715275,
      "loss": 1.1037,
      "step": 2187
    },
    {
      "epoch": 0.1577164275931666,
      "grad_norm": 0.22047922829871167,
      "learning_rate": 0.0001915728896847567,
      "loss": 1.1667,
      "step": 2188
    },
    {
      "epoch": 0.15778851005550348,
      "grad_norm": 0.14107556554331788,
      "learning_rate": 0.00019156350638081896,
      "loss": 1.0493,
      "step": 2189
    },
    {
      "epoch": 0.1578605925178404,
      "grad_norm": 0.20547949343174707,
      "learning_rate": 0.00019155411808585094,
      "loss": 1.0899,
      "step": 2190
    },
    {
      "epoch": 0.15793267498017732,
      "grad_norm": 0.12990838859151838,
      "learning_rate": 0.00019154472480036443,
      "loss": 1.1198,
      "step": 2191
    },
    {
      "epoch": 0.15800475744251424,
      "grad_norm": 0.35305550909946404,
      "learning_rate": 0.00019153532652487143,
      "loss": 1.1187,
      "step": 2192
    },
    {
      "epoch": 0.15807683990485116,
      "grad_norm": 0.17669597736820847,
      "learning_rate": 0.00019152592325988424,
      "loss": 1.1263,
      "step": 2193
    },
    {
      "epoch": 0.15814892236718806,
      "grad_norm": 0.11299660178566417,
      "learning_rate": 0.00019151651500591544,
      "loss": 1.1267,
      "step": 2194
    },
    {
      "epoch": 0.15822100482952497,
      "grad_norm": 0.11587748076937586,
      "learning_rate": 0.00019150710176347782,
      "loss": 1.1249,
      "step": 2195
    },
    {
      "epoch": 0.1582930872918619,
      "grad_norm": 0.10126206685079583,
      "learning_rate": 0.00019149768353308452,
      "loss": 1.1142,
      "step": 2196
    },
    {
      "epoch": 0.15836516975419881,
      "grad_norm": 0.11818403477157038,
      "learning_rate": 0.0001914882603152489,
      "loss": 1.1174,
      "step": 2197
    },
    {
      "epoch": 0.1584372522165357,
      "grad_norm": 0.13328215576526836,
      "learning_rate": 0.00019147883211048457,
      "loss": 1.0848,
      "step": 2198
    },
    {
      "epoch": 0.15850933467887263,
      "grad_norm": 0.32610120648191515,
      "learning_rate": 0.00019146939891930557,
      "loss": 1.1288,
      "step": 2199
    },
    {
      "epoch": 0.15858141714120955,
      "grad_norm": 0.14517992756463202,
      "learning_rate": 0.00019145996074222595,
      "loss": 1.1298,
      "step": 2200
    },
    {
      "epoch": 0.15865349960354647,
      "grad_norm": 0.13432436717534393,
      "learning_rate": 0.0001914505175797603,
      "loss": 1.1977,
      "step": 2201
    },
    {
      "epoch": 0.15872558206588336,
      "grad_norm": 0.16514776822057753,
      "learning_rate": 0.0001914410694324233,
      "loss": 1.1767,
      "step": 2202
    },
    {
      "epoch": 0.15879766452822028,
      "grad_norm": 2.684188887118528,
      "learning_rate": 0.00019143161630072995,
      "loss": 1.0966,
      "step": 2203
    },
    {
      "epoch": 0.1588697469905572,
      "grad_norm": 0.09878459882645098,
      "learning_rate": 0.00019142215818519553,
      "loss": 1.0754,
      "step": 2204
    },
    {
      "epoch": 0.15894182945289412,
      "grad_norm": 0.1265360998018131,
      "learning_rate": 0.00019141269508633562,
      "loss": 1.1425,
      "step": 2205
    },
    {
      "epoch": 0.15901391191523104,
      "grad_norm": 0.22140950557233968,
      "learning_rate": 0.000191403227004666,
      "loss": 1.1396,
      "step": 2206
    },
    {
      "epoch": 0.15908599437756793,
      "grad_norm": 0.34154780465360085,
      "learning_rate": 0.00019139375394070283,
      "loss": 1.0668,
      "step": 2207
    },
    {
      "epoch": 0.15915807683990485,
      "grad_norm": 0.16514631064590757,
      "learning_rate": 0.0001913842758949624,
      "loss": 1.1148,
      "step": 2208
    },
    {
      "epoch": 0.15923015930224177,
      "grad_norm": 0.10438415070866194,
      "learning_rate": 0.0001913747928679614,
      "loss": 1.1251,
      "step": 2209
    },
    {
      "epoch": 0.1593022417645787,
      "grad_norm": 0.1271092349402577,
      "learning_rate": 0.00019136530486021672,
      "loss": 1.1565,
      "step": 2210
    },
    {
      "epoch": 0.15937432422691558,
      "grad_norm": 0.12906930442378106,
      "learning_rate": 0.00019135581187224555,
      "loss": 1.0596,
      "step": 2211
    },
    {
      "epoch": 0.1594464066892525,
      "grad_norm": 0.23291007026171273,
      "learning_rate": 0.00019134631390456535,
      "loss": 1.1362,
      "step": 2212
    },
    {
      "epoch": 0.15951848915158942,
      "grad_norm": 0.19537195492102527,
      "learning_rate": 0.00019133681095769386,
      "loss": 1.1149,
      "step": 2213
    },
    {
      "epoch": 0.15959057161392634,
      "grad_norm": 0.11497335382669435,
      "learning_rate": 0.000191327303032149,
      "loss": 1.1283,
      "step": 2214
    },
    {
      "epoch": 0.15966265407626323,
      "grad_norm": 0.1957971886715832,
      "learning_rate": 0.00019131779012844912,
      "loss": 1.0869,
      "step": 2215
    },
    {
      "epoch": 0.15973473653860015,
      "grad_norm": 0.16058033277051473,
      "learning_rate": 0.00019130827224711272,
      "loss": 1.0807,
      "step": 2216
    },
    {
      "epoch": 0.15980681900093707,
      "grad_norm": 0.21288042947762395,
      "learning_rate": 0.00019129874938865861,
      "loss": 1.0068,
      "step": 2217
    },
    {
      "epoch": 0.159878901463274,
      "grad_norm": 0.1543798256144534,
      "learning_rate": 0.00019128922155360587,
      "loss": 1.074,
      "step": 2218
    },
    {
      "epoch": 0.1599509839256109,
      "grad_norm": 0.09329109618148927,
      "learning_rate": 0.00019127968874247387,
      "loss": 1.0738,
      "step": 2219
    },
    {
      "epoch": 0.1600230663879478,
      "grad_norm": 0.11359951558883793,
      "learning_rate": 0.00019127015095578222,
      "loss": 1.1608,
      "step": 2220
    },
    {
      "epoch": 0.16009514885028472,
      "grad_norm": 0.12729728481425281,
      "learning_rate": 0.0001912606081940508,
      "loss": 1.0342,
      "step": 2221
    },
    {
      "epoch": 0.16016723131262164,
      "grad_norm": 0.15293756707869943,
      "learning_rate": 0.00019125106045779982,
      "loss": 1.0201,
      "step": 2222
    },
    {
      "epoch": 0.16023931377495856,
      "grad_norm": 0.1294911286251305,
      "learning_rate": 0.0001912415077475497,
      "loss": 1.16,
      "step": 2223
    },
    {
      "epoch": 0.16031139623729546,
      "grad_norm": 0.11562112484129758,
      "learning_rate": 0.0001912319500638211,
      "loss": 1.1124,
      "step": 2224
    },
    {
      "epoch": 0.16038347869963238,
      "grad_norm": 0.12086438871297851,
      "learning_rate": 0.00019122238740713505,
      "loss": 1.0261,
      "step": 2225
    },
    {
      "epoch": 0.1604555611619693,
      "grad_norm": 0.11417586262533018,
      "learning_rate": 0.0001912128197780128,
      "loss": 1.132,
      "step": 2226
    },
    {
      "epoch": 0.16052764362430622,
      "grad_norm": 0.15315912784718036,
      "learning_rate": 0.00019120324717697583,
      "loss": 1.2069,
      "step": 2227
    },
    {
      "epoch": 0.1605997260866431,
      "grad_norm": 0.1148747372212876,
      "learning_rate": 0.00019119366960454598,
      "loss": 1.1219,
      "step": 2228
    },
    {
      "epoch": 0.16067180854898003,
      "grad_norm": 0.11420209034368134,
      "learning_rate": 0.00019118408706124532,
      "loss": 1.0594,
      "step": 2229
    },
    {
      "epoch": 0.16074389101131695,
      "grad_norm": 0.6410575276758924,
      "learning_rate": 0.00019117449954759612,
      "loss": 1.118,
      "step": 2230
    },
    {
      "epoch": 0.16081597347365387,
      "grad_norm": 0.6954709726246836,
      "learning_rate": 0.00019116490706412104,
      "loss": 1.1279,
      "step": 2231
    },
    {
      "epoch": 0.1608880559359908,
      "grad_norm": 1.6022757464530843,
      "learning_rate": 0.00019115530961134295,
      "loss": 1.129,
      "step": 2232
    },
    {
      "epoch": 0.16096013839832768,
      "grad_norm": 0.10842687836104623,
      "learning_rate": 0.00019114570718978496,
      "loss": 1.0605,
      "step": 2233
    },
    {
      "epoch": 0.1610322208606646,
      "grad_norm": 1.7293096801354582,
      "learning_rate": 0.00019113609979997054,
      "loss": 1.1269,
      "step": 2234
    },
    {
      "epoch": 0.16110430332300152,
      "grad_norm": 0.20084670364034046,
      "learning_rate": 0.00019112648744242335,
      "loss": 1.1221,
      "step": 2235
    },
    {
      "epoch": 0.16117638578533844,
      "grad_norm": 0.13374894938803272,
      "learning_rate": 0.00019111687011766733,
      "loss": 1.1101,
      "step": 2236
    },
    {
      "epoch": 0.16124846824767533,
      "grad_norm": 0.1276168952119773,
      "learning_rate": 0.00019110724782622676,
      "loss": 1.0655,
      "step": 2237
    },
    {
      "epoch": 0.16132055071001225,
      "grad_norm": 0.13253065284102714,
      "learning_rate": 0.00019109762056862613,
      "loss": 1.072,
      "step": 2238
    },
    {
      "epoch": 0.16139263317234917,
      "grad_norm": 0.12320470265826543,
      "learning_rate": 0.00019108798834539017,
      "loss": 1.02,
      "step": 2239
    },
    {
      "epoch": 0.1614647156346861,
      "grad_norm": 0.10566741450377966,
      "learning_rate": 0.00019107835115704394,
      "loss": 1.1365,
      "step": 2240
    },
    {
      "epoch": 0.16153679809702298,
      "grad_norm": 0.1275022008396782,
      "learning_rate": 0.00019106870900411277,
      "loss": 1.0981,
      "step": 2241
    },
    {
      "epoch": 0.1616088805593599,
      "grad_norm": 0.22076417633019083,
      "learning_rate": 0.00019105906188712223,
      "loss": 1.1263,
      "step": 2242
    },
    {
      "epoch": 0.16168096302169682,
      "grad_norm": 0.1127541493018923,
      "learning_rate": 0.0001910494098065982,
      "loss": 1.0891,
      "step": 2243
    },
    {
      "epoch": 0.16175304548403374,
      "grad_norm": 0.12275996745582902,
      "learning_rate": 0.00019103975276306678,
      "loss": 1.0987,
      "step": 2244
    },
    {
      "epoch": 0.16182512794637066,
      "grad_norm": 2.4657388579762047,
      "learning_rate": 0.00019103009075705437,
      "loss": 1.0951,
      "step": 2245
    },
    {
      "epoch": 0.16189721040870755,
      "grad_norm": 0.3869286472320543,
      "learning_rate": 0.00019102042378908761,
      "loss": 1.1166,
      "step": 2246
    },
    {
      "epoch": 0.16196929287104447,
      "grad_norm": 0.13770725709491485,
      "learning_rate": 0.00019101075185969348,
      "loss": 1.0754,
      "step": 2247
    },
    {
      "epoch": 0.1620413753333814,
      "grad_norm": 0.2838097852880583,
      "learning_rate": 0.00019100107496939917,
      "loss": 1.1268,
      "step": 2248
    },
    {
      "epoch": 0.1621134577957183,
      "grad_norm": 2.7442209442158405,
      "learning_rate": 0.00019099139311873215,
      "loss": 1.1516,
      "step": 2249
    },
    {
      "epoch": 0.1621855402580552,
      "grad_norm": 0.1614894969730873,
      "learning_rate": 0.0001909817063082202,
      "loss": 1.1691,
      "step": 2250
    },
    {
      "epoch": 0.16225762272039213,
      "grad_norm": 0.13471746099414178,
      "learning_rate": 0.00019097201453839127,
      "loss": 1.0749,
      "step": 2251
    },
    {
      "epoch": 0.16232970518272904,
      "grad_norm": 0.14076884884453464,
      "learning_rate": 0.0001909623178097737,
      "loss": 1.0808,
      "step": 2252
    },
    {
      "epoch": 0.16240178764506596,
      "grad_norm": 0.120490492327519,
      "learning_rate": 0.00019095261612289607,
      "loss": 1.0975,
      "step": 2253
    },
    {
      "epoch": 0.16247387010740286,
      "grad_norm": 0.3752238871977062,
      "learning_rate": 0.00019094290947828713,
      "loss": 1.1498,
      "step": 2254
    },
    {
      "epoch": 0.16254595256973978,
      "grad_norm": 0.11417703582166382,
      "learning_rate": 0.00019093319787647604,
      "loss": 1.0293,
      "step": 2255
    },
    {
      "epoch": 0.1626180350320767,
      "grad_norm": 0.1592383112103573,
      "learning_rate": 0.00019092348131799215,
      "loss": 1.2283,
      "step": 2256
    },
    {
      "epoch": 0.16269011749441362,
      "grad_norm": 1.0303492830868062,
      "learning_rate": 0.00019091375980336513,
      "loss": 1.1257,
      "step": 2257
    },
    {
      "epoch": 0.16276219995675054,
      "grad_norm": 0.25804328544620775,
      "learning_rate": 0.00019090403333312485,
      "loss": 1.2142,
      "step": 2258
    },
    {
      "epoch": 0.16283428241908743,
      "grad_norm": 0.273365717484456,
      "learning_rate": 0.00019089430190780149,
      "loss": 1.1791,
      "step": 2259
    },
    {
      "epoch": 0.16290636488142435,
      "grad_norm": 0.15150682684042882,
      "learning_rate": 0.00019088456552792554,
      "loss": 1.0978,
      "step": 2260
    },
    {
      "epoch": 0.16297844734376127,
      "grad_norm": 0.6549260177996965,
      "learning_rate": 0.00019087482419402764,
      "loss": 1.0124,
      "step": 2261
    },
    {
      "epoch": 0.1630505298060982,
      "grad_norm": 0.20831377709845975,
      "learning_rate": 0.00019086507790663885,
      "loss": 1.0993,
      "step": 2262
    },
    {
      "epoch": 0.16312261226843508,
      "grad_norm": 0.0964894883385884,
      "learning_rate": 0.00019085532666629045,
      "loss": 1.0495,
      "step": 2263
    },
    {
      "epoch": 0.163194694730772,
      "grad_norm": 0.12687926500527852,
      "learning_rate": 0.0001908455704735139,
      "loss": 1.1184,
      "step": 2264
    },
    {
      "epoch": 0.16326677719310892,
      "grad_norm": 0.39480445004396725,
      "learning_rate": 0.00019083580932884104,
      "loss": 1.1215,
      "step": 2265
    },
    {
      "epoch": 0.16333885965544584,
      "grad_norm": 0.46028883551633926,
      "learning_rate": 0.00019082604323280393,
      "loss": 1.1138,
      "step": 2266
    },
    {
      "epoch": 0.16341094211778273,
      "grad_norm": 0.19323402738674741,
      "learning_rate": 0.0001908162721859349,
      "loss": 1.1103,
      "step": 2267
    },
    {
      "epoch": 0.16348302458011965,
      "grad_norm": 0.47248608128772607,
      "learning_rate": 0.0001908064961887666,
      "loss": 1.179,
      "step": 2268
    },
    {
      "epoch": 0.16355510704245657,
      "grad_norm": 0.21273911965403328,
      "learning_rate": 0.00019079671524183185,
      "loss": 1.1436,
      "step": 2269
    },
    {
      "epoch": 0.1636271895047935,
      "grad_norm": 0.5063472291389315,
      "learning_rate": 0.00019078692934566385,
      "loss": 1.0985,
      "step": 2270
    },
    {
      "epoch": 0.1636992719671304,
      "grad_norm": 0.10610785313480568,
      "learning_rate": 0.00019077713850079598,
      "loss": 1.0726,
      "step": 2271
    },
    {
      "epoch": 0.1637713544294673,
      "grad_norm": 0.42091435021899404,
      "learning_rate": 0.00019076734270776196,
      "loss": 1.078,
      "step": 2272
    },
    {
      "epoch": 0.16384343689180422,
      "grad_norm": 0.1526892841863034,
      "learning_rate": 0.00019075754196709572,
      "loss": 1.07,
      "step": 2273
    },
    {
      "epoch": 0.16391551935414114,
      "grad_norm": 0.29810758852591185,
      "learning_rate": 0.0001907477362793315,
      "loss": 1.049,
      "step": 2274
    },
    {
      "epoch": 0.16398760181647806,
      "grad_norm": 0.3285788098531825,
      "learning_rate": 0.0001907379256450038,
      "loss": 1.0691,
      "step": 2275
    },
    {
      "epoch": 0.16405968427881495,
      "grad_norm": 0.2988490560229674,
      "learning_rate": 0.00019072811006464743,
      "loss": 1.0837,
      "step": 2276
    },
    {
      "epoch": 0.16413176674115187,
      "grad_norm": 0.19442624223498528,
      "learning_rate": 0.00019071828953879733,
      "loss": 1.1054,
      "step": 2277
    },
    {
      "epoch": 0.1642038492034888,
      "grad_norm": 0.280872220386195,
      "learning_rate": 0.0001907084640679889,
      "loss": 1.1047,
      "step": 2278
    },
    {
      "epoch": 0.16427593166582571,
      "grad_norm": 0.24875646763935794,
      "learning_rate": 0.00019069863365275765,
      "loss": 1.1463,
      "step": 2279
    },
    {
      "epoch": 0.1643480141281626,
      "grad_norm": 0.12784237523396685,
      "learning_rate": 0.00019068879829363947,
      "loss": 1.1816,
      "step": 2280
    },
    {
      "epoch": 0.16442009659049953,
      "grad_norm": 0.3098496837749114,
      "learning_rate": 0.00019067895799117045,
      "loss": 1.1118,
      "step": 2281
    },
    {
      "epoch": 0.16449217905283645,
      "grad_norm": 0.10959663218021874,
      "learning_rate": 0.00019066911274588698,
      "loss": 1.1968,
      "step": 2282
    },
    {
      "epoch": 0.16456426151517337,
      "grad_norm": 0.15545839363311956,
      "learning_rate": 0.00019065926255832573,
      "loss": 1.144,
      "step": 2283
    },
    {
      "epoch": 0.16463634397751029,
      "grad_norm": 0.20041212756159396,
      "learning_rate": 0.00019064940742902358,
      "loss": 1.1718,
      "step": 2284
    },
    {
      "epoch": 0.16470842643984718,
      "grad_norm": 0.11137826030735255,
      "learning_rate": 0.00019063954735851777,
      "loss": 1.1484,
      "step": 2285
    },
    {
      "epoch": 0.1647805089021841,
      "grad_norm": 0.1529586217191371,
      "learning_rate": 0.00019062968234734577,
      "loss": 1.17,
      "step": 2286
    },
    {
      "epoch": 0.16485259136452102,
      "grad_norm": 0.12026331599895813,
      "learning_rate": 0.00019061981239604524,
      "loss": 1.1234,
      "step": 2287
    },
    {
      "epoch": 0.16492467382685794,
      "grad_norm": 0.15705683721516783,
      "learning_rate": 0.00019060993750515427,
      "loss": 1.0984,
      "step": 2288
    },
    {
      "epoch": 0.16499675628919483,
      "grad_norm": 0.12253864544244385,
      "learning_rate": 0.00019060005767521104,
      "loss": 1.1428,
      "step": 2289
    },
    {
      "epoch": 0.16506883875153175,
      "grad_norm": 0.1807207986634044,
      "learning_rate": 0.00019059017290675416,
      "loss": 1.0574,
      "step": 2290
    },
    {
      "epoch": 0.16514092121386867,
      "grad_norm": 0.1341174971020015,
      "learning_rate": 0.0001905802832003224,
      "loss": 1.0632,
      "step": 2291
    },
    {
      "epoch": 0.1652130036762056,
      "grad_norm": 0.13537210973484712,
      "learning_rate": 0.00019057038855645487,
      "loss": 1.0657,
      "step": 2292
    },
    {
      "epoch": 0.16528508613854248,
      "grad_norm": 0.23859431659929622,
      "learning_rate": 0.00019056048897569093,
      "loss": 1.0927,
      "step": 2293
    },
    {
      "epoch": 0.1653571686008794,
      "grad_norm": 0.1873378396378719,
      "learning_rate": 0.00019055058445857013,
      "loss": 1.1344,
      "step": 2294
    },
    {
      "epoch": 0.16542925106321632,
      "grad_norm": 0.12284507337063749,
      "learning_rate": 0.00019054067500563236,
      "loss": 1.1095,
      "step": 2295
    },
    {
      "epoch": 0.16550133352555324,
      "grad_norm": 0.137092206365341,
      "learning_rate": 0.00019053076061741783,
      "loss": 1.123,
      "step": 2296
    },
    {
      "epoch": 0.16557341598789016,
      "grad_norm": 0.3584747732722756,
      "learning_rate": 0.00019052084129446694,
      "loss": 1.1046,
      "step": 2297
    },
    {
      "epoch": 0.16564549845022705,
      "grad_norm": 0.12161992339412991,
      "learning_rate": 0.00019051091703732037,
      "loss": 1.1352,
      "step": 2298
    },
    {
      "epoch": 0.16571758091256397,
      "grad_norm": 0.12280471428694373,
      "learning_rate": 0.0001905009878465191,
      "loss": 1.0534,
      "step": 2299
    },
    {
      "epoch": 0.1657896633749009,
      "grad_norm": 0.37900078827200223,
      "learning_rate": 0.00019049105372260432,
      "loss": 1.0966,
      "step": 2300
    },
    {
      "epoch": 0.1658617458372378,
      "grad_norm": 0.28304806271028027,
      "learning_rate": 0.00019048111466611758,
      "loss": 1.1547,
      "step": 2301
    },
    {
      "epoch": 0.1659338282995747,
      "grad_norm": 0.10887772441958271,
      "learning_rate": 0.0001904711706776006,
      "loss": 1.0587,
      "step": 2302
    },
    {
      "epoch": 0.16600591076191162,
      "grad_norm": 0.1095640979865937,
      "learning_rate": 0.00019046122175759547,
      "loss": 1.0648,
      "step": 2303
    },
    {
      "epoch": 0.16607799322424854,
      "grad_norm": 0.12904294240083386,
      "learning_rate": 0.00019045126790664444,
      "loss": 1.1636,
      "step": 2304
    },
    {
      "epoch": 0.16615007568658546,
      "grad_norm": 0.1140323646183745,
      "learning_rate": 0.00019044130912529013,
      "loss": 1.1576,
      "step": 2305
    },
    {
      "epoch": 0.16622215814892236,
      "grad_norm": 0.4177540319798719,
      "learning_rate": 0.00019043134541407532,
      "loss": 1.1602,
      "step": 2306
    },
    {
      "epoch": 0.16629424061125928,
      "grad_norm": 0.12368994426558487,
      "learning_rate": 0.00019042137677354322,
      "loss": 1.179,
      "step": 2307
    },
    {
      "epoch": 0.1663663230735962,
      "grad_norm": 0.22808148458759486,
      "learning_rate": 0.00019041140320423712,
      "loss": 1.1947,
      "step": 2308
    },
    {
      "epoch": 0.16643840553593311,
      "grad_norm": 0.26687178542624357,
      "learning_rate": 0.0001904014247067007,
      "loss": 1.1242,
      "step": 2309
    },
    {
      "epoch": 0.16651048799827003,
      "grad_norm": 0.12601664494602563,
      "learning_rate": 0.00019039144128147787,
      "loss": 1.1703,
      "step": 2310
    },
    {
      "epoch": 0.16658257046060693,
      "grad_norm": 0.3055082446026757,
      "learning_rate": 0.00019038145292911285,
      "loss": 1.1199,
      "step": 2311
    },
    {
      "epoch": 0.16665465292294385,
      "grad_norm": 0.11563827088631881,
      "learning_rate": 0.00019037145965015005,
      "loss": 1.1491,
      "step": 2312
    },
    {
      "epoch": 0.16672673538528077,
      "grad_norm": 0.14662951395654789,
      "learning_rate": 0.00019036146144513425,
      "loss": 1.1689,
      "step": 2313
    },
    {
      "epoch": 0.1667988178476177,
      "grad_norm": 0.1486287923971109,
      "learning_rate": 0.0001903514583146104,
      "loss": 1.1342,
      "step": 2314
    },
    {
      "epoch": 0.16687090030995458,
      "grad_norm": 0.19756519966889097,
      "learning_rate": 0.0001903414502591237,
      "loss": 1.1397,
      "step": 2315
    },
    {
      "epoch": 0.1669429827722915,
      "grad_norm": 0.127105950516049,
      "learning_rate": 0.0001903314372792198,
      "loss": 1.1227,
      "step": 2316
    },
    {
      "epoch": 0.16701506523462842,
      "grad_norm": 0.3156796363320524,
      "learning_rate": 0.00019032141937544444,
      "loss": 0.963,
      "step": 2317
    },
    {
      "epoch": 0.16708714769696534,
      "grad_norm": 0.10951760142647254,
      "learning_rate": 0.00019031139654834367,
      "loss": 1.1031,
      "step": 2318
    },
    {
      "epoch": 0.16715923015930223,
      "grad_norm": 0.14837355133051564,
      "learning_rate": 0.00019030136879846383,
      "loss": 1.153,
      "step": 2319
    },
    {
      "epoch": 0.16723131262163915,
      "grad_norm": 0.19896754129293995,
      "learning_rate": 0.00019029133612635154,
      "loss": 1.1708,
      "step": 2320
    },
    {
      "epoch": 0.16730339508397607,
      "grad_norm": 0.1117764242101092,
      "learning_rate": 0.00019028129853255367,
      "loss": 1.0824,
      "step": 2321
    },
    {
      "epoch": 0.167375477546313,
      "grad_norm": 0.17277989034957644,
      "learning_rate": 0.00019027125601761734,
      "loss": 1.054,
      "step": 2322
    },
    {
      "epoch": 0.16744756000864988,
      "grad_norm": 0.1543259714515994,
      "learning_rate": 0.00019026120858208997,
      "loss": 1.1554,
      "step": 2323
    },
    {
      "epoch": 0.1675196424709868,
      "grad_norm": 0.24419622340874497,
      "learning_rate": 0.00019025115622651922,
      "loss": 1.1217,
      "step": 2324
    },
    {
      "epoch": 0.16759172493332372,
      "grad_norm": 0.12865239638188175,
      "learning_rate": 0.00019024109895145308,
      "loss": 1.0367,
      "step": 2325
    },
    {
      "epoch": 0.16766380739566064,
      "grad_norm": 0.1439787164496055,
      "learning_rate": 0.00019023103675743972,
      "loss": 1.0853,
      "step": 2326
    },
    {
      "epoch": 0.16773588985799756,
      "grad_norm": 0.3142982116227015,
      "learning_rate": 0.0001902209696450276,
      "loss": 1.0629,
      "step": 2327
    },
    {
      "epoch": 0.16780797232033445,
      "grad_norm": 0.13029525210740095,
      "learning_rate": 0.0001902108976147655,
      "loss": 1.1104,
      "step": 2328
    },
    {
      "epoch": 0.16788005478267137,
      "grad_norm": 0.1367411329610758,
      "learning_rate": 0.00019020082066720245,
      "loss": 1.046,
      "step": 2329
    },
    {
      "epoch": 0.1679521372450083,
      "grad_norm": 0.2001868309219134,
      "learning_rate": 0.0001901907388028877,
      "loss": 1.1443,
      "step": 2330
    },
    {
      "epoch": 0.1680242197073452,
      "grad_norm": 1.7920585332204846,
      "learning_rate": 0.00019018065202237083,
      "loss": 1.0871,
      "step": 2331
    },
    {
      "epoch": 0.1680963021696821,
      "grad_norm": 0.11915712202303966,
      "learning_rate": 0.00019017056032620162,
      "loss": 1.0738,
      "step": 2332
    },
    {
      "epoch": 0.16816838463201902,
      "grad_norm": 0.10943503264022537,
      "learning_rate": 0.0001901604637149302,
      "loss": 1.1379,
      "step": 2333
    },
    {
      "epoch": 0.16824046709435594,
      "grad_norm": 0.12116010697992018,
      "learning_rate": 0.0001901503621891069,
      "loss": 1.0138,
      "step": 2334
    },
    {
      "epoch": 0.16831254955669286,
      "grad_norm": 0.14168587115741715,
      "learning_rate": 0.00019014025574928237,
      "loss": 1.1811,
      "step": 2335
    },
    {
      "epoch": 0.16838463201902976,
      "grad_norm": 0.11485092719747697,
      "learning_rate": 0.00019013014439600746,
      "loss": 1.1109,
      "step": 2336
    },
    {
      "epoch": 0.16845671448136668,
      "grad_norm": 0.11378914891210708,
      "learning_rate": 0.00019012002812983337,
      "loss": 1.0704,
      "step": 2337
    },
    {
      "epoch": 0.1685287969437036,
      "grad_norm": 0.11982786305349163,
      "learning_rate": 0.00019010990695131152,
      "loss": 1.1098,
      "step": 2338
    },
    {
      "epoch": 0.16860087940604052,
      "grad_norm": 0.1020379032119147,
      "learning_rate": 0.0001900997808609936,
      "loss": 1.114,
      "step": 2339
    },
    {
      "epoch": 0.16867296186837744,
      "grad_norm": 0.12843028096820652,
      "learning_rate": 0.00019008964985943151,
      "loss": 1.0562,
      "step": 2340
    },
    {
      "epoch": 0.16874504433071433,
      "grad_norm": 0.10976072650827587,
      "learning_rate": 0.0001900795139471776,
      "loss": 1.082,
      "step": 2341
    },
    {
      "epoch": 0.16881712679305125,
      "grad_norm": 0.18183200896923868,
      "learning_rate": 0.00019006937312478428,
      "loss": 1.1055,
      "step": 2342
    },
    {
      "epoch": 0.16888920925538817,
      "grad_norm": 0.11422812524009376,
      "learning_rate": 0.00019005922739280434,
      "loss": 1.0424,
      "step": 2343
    },
    {
      "epoch": 0.1689612917177251,
      "grad_norm": 0.12328296317800423,
      "learning_rate": 0.00019004907675179082,
      "loss": 1.1076,
      "step": 2344
    },
    {
      "epoch": 0.16903337418006198,
      "grad_norm": 0.12944704859873205,
      "learning_rate": 0.000190038921202297,
      "loss": 1.0953,
      "step": 2345
    },
    {
      "epoch": 0.1691054566423989,
      "grad_norm": 0.22662734910520999,
      "learning_rate": 0.00019002876074487648,
      "loss": 1.0405,
      "step": 2346
    },
    {
      "epoch": 0.16917753910473582,
      "grad_norm": 0.15288456600337447,
      "learning_rate": 0.0001900185953800831,
      "loss": 1.0884,
      "step": 2347
    },
    {
      "epoch": 0.16924962156707274,
      "grad_norm": 0.11387286637398851,
      "learning_rate": 0.0001900084251084709,
      "loss": 1.1455,
      "step": 2348
    },
    {
      "epoch": 0.16932170402940963,
      "grad_norm": 0.11569167620757412,
      "learning_rate": 0.0001899982499305943,
      "loss": 1.1061,
      "step": 2349
    },
    {
      "epoch": 0.16939378649174655,
      "grad_norm": 0.19129991257105913,
      "learning_rate": 0.00018998806984700796,
      "loss": 1.1059,
      "step": 2350
    },
    {
      "epoch": 0.16946586895408347,
      "grad_norm": 0.2186360657315151,
      "learning_rate": 0.00018997788485826674,
      "loss": 0.9912,
      "step": 2351
    },
    {
      "epoch": 0.1695379514164204,
      "grad_norm": 0.1306575011844316,
      "learning_rate": 0.0001899676949649258,
      "loss": 1.1187,
      "step": 2352
    },
    {
      "epoch": 0.1696100338787573,
      "grad_norm": 0.09981047652505455,
      "learning_rate": 0.00018995750016754067,
      "loss": 1.0517,
      "step": 2353
    },
    {
      "epoch": 0.1696821163410942,
      "grad_norm": 0.11879911054252118,
      "learning_rate": 0.00018994730046666694,
      "loss": 1.055,
      "step": 2354
    },
    {
      "epoch": 0.16975419880343112,
      "grad_norm": 0.11292373384669908,
      "learning_rate": 0.0001899370958628607,
      "loss": 1.0233,
      "step": 2355
    },
    {
      "epoch": 0.16982628126576804,
      "grad_norm": 0.11121562213630119,
      "learning_rate": 0.00018992688635667811,
      "loss": 1.1338,
      "step": 2356
    },
    {
      "epoch": 0.16989836372810496,
      "grad_norm": 0.10652696793726664,
      "learning_rate": 0.00018991667194867567,
      "loss": 1.0648,
      "step": 2357
    },
    {
      "epoch": 0.16997044619044185,
      "grad_norm": 0.17606508245341,
      "learning_rate": 0.00018990645263941021,
      "loss": 1.1348,
      "step": 2358
    },
    {
      "epoch": 0.17004252865277877,
      "grad_norm": 1.3317986198407898,
      "learning_rate": 0.0001898962284294388,
      "loss": 1.1129,
      "step": 2359
    },
    {
      "epoch": 0.1701146111151157,
      "grad_norm": 0.16932610043974353,
      "learning_rate": 0.00018988599931931866,
      "loss": 1.0431,
      "step": 2360
    },
    {
      "epoch": 0.1701866935774526,
      "grad_norm": 0.23390780440084308,
      "learning_rate": 0.00018987576530960743,
      "loss": 1.1553,
      "step": 2361
    },
    {
      "epoch": 0.1702587760397895,
      "grad_norm": 1.255023971807726,
      "learning_rate": 0.00018986552640086295,
      "loss": 1.0676,
      "step": 2362
    },
    {
      "epoch": 0.17033085850212643,
      "grad_norm": 0.15423305613760144,
      "learning_rate": 0.0001898552825936433,
      "loss": 1.0672,
      "step": 2363
    },
    {
      "epoch": 0.17040294096446335,
      "grad_norm": 0.12758105725716723,
      "learning_rate": 0.00018984503388850686,
      "loss": 1.1357,
      "step": 2364
    },
    {
      "epoch": 0.17047502342680027,
      "grad_norm": 0.11944748271666433,
      "learning_rate": 0.00018983478028601234,
      "loss": 1.1581,
      "step": 2365
    },
    {
      "epoch": 0.17054710588913718,
      "grad_norm": 0.12976674338690697,
      "learning_rate": 0.0001898245217867186,
      "loss": 1.1353,
      "step": 2366
    },
    {
      "epoch": 0.17061918835147408,
      "grad_norm": 0.15906193396804413,
      "learning_rate": 0.00018981425839118484,
      "loss": 1.0133,
      "step": 2367
    },
    {
      "epoch": 0.170691270813811,
      "grad_norm": 0.12788874384014137,
      "learning_rate": 0.00018980399009997046,
      "loss": 1.2239,
      "step": 2368
    },
    {
      "epoch": 0.17076335327614792,
      "grad_norm": 0.13662069896754522,
      "learning_rate": 0.00018979371691363523,
      "loss": 1.0654,
      "step": 2369
    },
    {
      "epoch": 0.17083543573848484,
      "grad_norm": 0.1440381858084433,
      "learning_rate": 0.0001897834388327391,
      "loss": 1.1033,
      "step": 2370
    },
    {
      "epoch": 0.17090751820082173,
      "grad_norm": 0.2628717962573232,
      "learning_rate": 0.00018977315585784235,
      "loss": 1.0832,
      "step": 2371
    },
    {
      "epoch": 0.17097960066315865,
      "grad_norm": 0.25487329916982343,
      "learning_rate": 0.00018976286798950547,
      "loss": 1.0569,
      "step": 2372
    },
    {
      "epoch": 0.17105168312549557,
      "grad_norm": 0.12989239702437366,
      "learning_rate": 0.0001897525752282892,
      "loss": 1.0198,
      "step": 2373
    },
    {
      "epoch": 0.1711237655878325,
      "grad_norm": 0.12960235210270782,
      "learning_rate": 0.0001897422775747547,
      "loss": 1.1055,
      "step": 2374
    },
    {
      "epoch": 0.17119584805016938,
      "grad_norm": 0.22273809681414827,
      "learning_rate": 0.00018973197502946315,
      "loss": 1.0933,
      "step": 2375
    },
    {
      "epoch": 0.1712679305125063,
      "grad_norm": 0.19194459851367893,
      "learning_rate": 0.00018972166759297622,
      "loss": 1.0451,
      "step": 2376
    },
    {
      "epoch": 0.17134001297484322,
      "grad_norm": 0.3073971828803981,
      "learning_rate": 0.00018971135526585573,
      "loss": 1.147,
      "step": 2377
    },
    {
      "epoch": 0.17141209543718014,
      "grad_norm": 0.14900636562068667,
      "learning_rate": 0.00018970103804866382,
      "loss": 1.0725,
      "step": 2378
    },
    {
      "epoch": 0.17148417789951706,
      "grad_norm": 0.22698318461346165,
      "learning_rate": 0.00018969071594196284,
      "loss": 1.0682,
      "step": 2379
    },
    {
      "epoch": 0.17155626036185395,
      "grad_norm": 0.14196853441015939,
      "learning_rate": 0.00018968038894631544,
      "loss": 1.0264,
      "step": 2380
    },
    {
      "epoch": 0.17162834282419087,
      "grad_norm": 0.20472698455037888,
      "learning_rate": 0.00018967005706228452,
      "loss": 1.1215,
      "step": 2381
    },
    {
      "epoch": 0.1717004252865278,
      "grad_norm": 0.6114690670083864,
      "learning_rate": 0.00018965972029043332,
      "loss": 1.1854,
      "step": 2382
    },
    {
      "epoch": 0.1717725077488647,
      "grad_norm": 0.20883822847606537,
      "learning_rate": 0.0001896493786313252,
      "loss": 1.1044,
      "step": 2383
    },
    {
      "epoch": 0.1718445902112016,
      "grad_norm": 0.11601639009308784,
      "learning_rate": 0.00018963903208552393,
      "loss": 1.0686,
      "step": 2384
    },
    {
      "epoch": 0.17191667267353852,
      "grad_norm": 0.31495554886291205,
      "learning_rate": 0.0001896286806535935,
      "loss": 1.0661,
      "step": 2385
    },
    {
      "epoch": 0.17198875513587544,
      "grad_norm": 0.2715330176120864,
      "learning_rate": 0.0001896183243360981,
      "loss": 1.1144,
      "step": 2386
    },
    {
      "epoch": 0.17206083759821236,
      "grad_norm": 0.1986802163330264,
      "learning_rate": 0.0001896079631336023,
      "loss": 1.1113,
      "step": 2387
    },
    {
      "epoch": 0.17213292006054925,
      "grad_norm": 0.1033857333040018,
      "learning_rate": 0.00018959759704667087,
      "loss": 1.0713,
      "step": 2388
    },
    {
      "epoch": 0.17220500252288617,
      "grad_norm": 0.12122295603845643,
      "learning_rate": 0.0001895872260758688,
      "loss": 1.1299,
      "step": 2389
    },
    {
      "epoch": 0.1722770849852231,
      "grad_norm": 0.12267028483169413,
      "learning_rate": 0.00018957685022176146,
      "loss": 1.0511,
      "step": 2390
    },
    {
      "epoch": 0.17234916744756001,
      "grad_norm": 0.10762288115556727,
      "learning_rate": 0.0001895664694849144,
      "loss": 1.1361,
      "step": 2391
    },
    {
      "epoch": 0.17242124990989693,
      "grad_norm": 0.11873046813225308,
      "learning_rate": 0.00018955608386589346,
      "loss": 1.142,
      "step": 2392
    },
    {
      "epoch": 0.17249333237223383,
      "grad_norm": 0.28413932558440547,
      "learning_rate": 0.00018954569336526478,
      "loss": 1.0912,
      "step": 2393
    },
    {
      "epoch": 0.17256541483457075,
      "grad_norm": 0.12209235625459376,
      "learning_rate": 0.0001895352979835947,
      "loss": 1.1411,
      "step": 2394
    },
    {
      "epoch": 0.17263749729690767,
      "grad_norm": 0.12909635308735418,
      "learning_rate": 0.00018952489772144988,
      "loss": 1.0912,
      "step": 2395
    },
    {
      "epoch": 0.17270957975924459,
      "grad_norm": 0.8755760426429058,
      "learning_rate": 0.0001895144925793972,
      "loss": 1.1519,
      "step": 2396
    },
    {
      "epoch": 0.17278166222158148,
      "grad_norm": 0.12438404729815658,
      "learning_rate": 0.00018950408255800387,
      "loss": 1.1072,
      "step": 2397
    },
    {
      "epoch": 0.1728537446839184,
      "grad_norm": 0.12736771986285256,
      "learning_rate": 0.00018949366765783732,
      "loss": 1.1056,
      "step": 2398
    },
    {
      "epoch": 0.17292582714625532,
      "grad_norm": 0.22124519893242092,
      "learning_rate": 0.00018948324787946525,
      "loss": 1.1822,
      "step": 2399
    },
    {
      "epoch": 0.17299790960859224,
      "grad_norm": 0.44966508300343194,
      "learning_rate": 0.00018947282322345562,
      "loss": 1.1189,
      "step": 2400
    },
    {
      "epoch": 0.17306999207092913,
      "grad_norm": 0.11015031023930301,
      "learning_rate": 0.00018946239369037668,
      "loss": 1.0498,
      "step": 2401
    },
    {
      "epoch": 0.17314207453326605,
      "grad_norm": 2.4701534438754766,
      "learning_rate": 0.00018945195928079693,
      "loss": 1.0778,
      "step": 2402
    },
    {
      "epoch": 0.17321415699560297,
      "grad_norm": 1.6895035389025266,
      "learning_rate": 0.00018944151999528516,
      "loss": 1.0448,
      "step": 2403
    },
    {
      "epoch": 0.1732862394579399,
      "grad_norm": 0.14029415928657882,
      "learning_rate": 0.00018943107583441036,
      "loss": 1.0975,
      "step": 2404
    },
    {
      "epoch": 0.1733583219202768,
      "grad_norm": 0.12087460949480135,
      "learning_rate": 0.00018942062679874186,
      "loss": 1.1217,
      "step": 2405
    },
    {
      "epoch": 0.1734304043826137,
      "grad_norm": 0.1362494302101068,
      "learning_rate": 0.0001894101728888492,
      "loss": 1.1571,
      "step": 2406
    },
    {
      "epoch": 0.17350248684495062,
      "grad_norm": 0.1289835109218869,
      "learning_rate": 0.00018939971410530224,
      "loss": 1.1267,
      "step": 2407
    },
    {
      "epoch": 0.17357456930728754,
      "grad_norm": 0.2102793758135076,
      "learning_rate": 0.0001893892504486711,
      "loss": 1.1832,
      "step": 2408
    },
    {
      "epoch": 0.17364665176962446,
      "grad_norm": 0.12412016358464804,
      "learning_rate": 0.00018937878191952606,
      "loss": 1.066,
      "step": 2409
    },
    {
      "epoch": 0.17371873423196135,
      "grad_norm": 0.17308647983931855,
      "learning_rate": 0.00018936830851843785,
      "loss": 1.1442,
      "step": 2410
    },
    {
      "epoch": 0.17379081669429827,
      "grad_norm": 1.3415901121999863,
      "learning_rate": 0.0001893578302459773,
      "loss": 1.0627,
      "step": 2411
    },
    {
      "epoch": 0.1738628991566352,
      "grad_norm": 0.14104826034538193,
      "learning_rate": 0.00018934734710271558,
      "loss": 1.0415,
      "step": 2412
    },
    {
      "epoch": 0.1739349816189721,
      "grad_norm": 0.12844067607249302,
      "learning_rate": 0.00018933685908922412,
      "loss": 1.0676,
      "step": 2413
    },
    {
      "epoch": 0.174007064081309,
      "grad_norm": 2.3731278145664687,
      "learning_rate": 0.0001893263662060746,
      "loss": 1.1166,
      "step": 2414
    },
    {
      "epoch": 0.17407914654364592,
      "grad_norm": 0.3971137290866068,
      "learning_rate": 0.00018931586845383897,
      "loss": 1.0164,
      "step": 2415
    },
    {
      "epoch": 0.17415122900598284,
      "grad_norm": 0.12380747797401626,
      "learning_rate": 0.0001893053658330895,
      "loss": 1.0428,
      "step": 2416
    },
    {
      "epoch": 0.17422331146831976,
      "grad_norm": 0.1459926037681207,
      "learning_rate": 0.00018929485834439863,
      "loss": 1.1012,
      "step": 2417
    },
    {
      "epoch": 0.17429539393065668,
      "grad_norm": 0.12111821266845396,
      "learning_rate": 0.00018928434598833912,
      "loss": 1.2016,
      "step": 2418
    },
    {
      "epoch": 0.17436747639299358,
      "grad_norm": 0.5346617912300138,
      "learning_rate": 0.000189273828765484,
      "loss": 1.2384,
      "step": 2419
    },
    {
      "epoch": 0.1744395588553305,
      "grad_norm": 0.25779586759642514,
      "learning_rate": 0.00018926330667640655,
      "loss": 1.0783,
      "step": 2420
    },
    {
      "epoch": 0.17451164131766742,
      "grad_norm": 0.1409150039813689,
      "learning_rate": 0.0001892527797216803,
      "loss": 1.116,
      "step": 2421
    },
    {
      "epoch": 0.17458372378000434,
      "grad_norm": 0.2641867946244299,
      "learning_rate": 0.0001892422479018791,
      "loss": 1.0793,
      "step": 2422
    },
    {
      "epoch": 0.17465580624234123,
      "grad_norm": 0.18600552394617803,
      "learning_rate": 0.00018923171121757698,
      "loss": 1.1277,
      "step": 2423
    },
    {
      "epoch": 0.17472788870467815,
      "grad_norm": 0.15584443515460653,
      "learning_rate": 0.00018922116966934834,
      "loss": 1.1256,
      "step": 2424
    },
    {
      "epoch": 0.17479997116701507,
      "grad_norm": 0.1270577444715749,
      "learning_rate": 0.00018921062325776773,
      "loss": 1.0098,
      "step": 2425
    },
    {
      "epoch": 0.174872053629352,
      "grad_norm": 0.8633393615724526,
      "learning_rate": 0.00018920007198341008,
      "loss": 1.1067,
      "step": 2426
    },
    {
      "epoch": 0.17494413609168888,
      "grad_norm": 0.11870554304964502,
      "learning_rate": 0.00018918951584685046,
      "loss": 1.0138,
      "step": 2427
    },
    {
      "epoch": 0.1750162185540258,
      "grad_norm": 0.1409523754572386,
      "learning_rate": 0.00018917895484866435,
      "loss": 1.1453,
      "step": 2428
    },
    {
      "epoch": 0.17508830101636272,
      "grad_norm": 0.2182620473083212,
      "learning_rate": 0.00018916838898942734,
      "loss": 1.0661,
      "step": 2429
    },
    {
      "epoch": 0.17516038347869964,
      "grad_norm": 1.0774264961668865,
      "learning_rate": 0.00018915781826971543,
      "loss": 1.0561,
      "step": 2430
    },
    {
      "epoch": 0.17523246594103656,
      "grad_norm": 0.33684198876826843,
      "learning_rate": 0.00018914724269010483,
      "loss": 1.0992,
      "step": 2431
    },
    {
      "epoch": 0.17530454840337345,
      "grad_norm": 0.3432479606227507,
      "learning_rate": 0.00018913666225117194,
      "loss": 1.1035,
      "step": 2432
    },
    {
      "epoch": 0.17537663086571037,
      "grad_norm": 0.16118010737769997,
      "learning_rate": 0.00018912607695349348,
      "loss": 0.9179,
      "step": 2433
    },
    {
      "epoch": 0.1754487133280473,
      "grad_norm": 0.14071775648995613,
      "learning_rate": 0.00018911548679764653,
      "loss": 1.148,
      "step": 2434
    },
    {
      "epoch": 0.1755207957903842,
      "grad_norm": 0.12709985670189153,
      "learning_rate": 0.0001891048917842083,
      "loss": 1.0472,
      "step": 2435
    },
    {
      "epoch": 0.1755928782527211,
      "grad_norm": 0.13682090339848105,
      "learning_rate": 0.0001890942919137563,
      "loss": 1.1457,
      "step": 2436
    },
    {
      "epoch": 0.17566496071505802,
      "grad_norm": 3.017250282805551,
      "learning_rate": 0.00018908368718686833,
      "loss": 1.0729,
      "step": 2437
    },
    {
      "epoch": 0.17573704317739494,
      "grad_norm": 0.6498268866142761,
      "learning_rate": 0.00018907307760412243,
      "loss": 1.0848,
      "step": 2438
    },
    {
      "epoch": 0.17580912563973186,
      "grad_norm": 0.24569806264684796,
      "learning_rate": 0.00018906246316609696,
      "loss": 1.0979,
      "step": 2439
    },
    {
      "epoch": 0.17588120810206875,
      "grad_norm": 0.1281019981563074,
      "learning_rate": 0.00018905184387337044,
      "loss": 1.0949,
      "step": 2440
    },
    {
      "epoch": 0.17595329056440567,
      "grad_norm": 0.18874318518882932,
      "learning_rate": 0.00018904121972652176,
      "loss": 1.1036,
      "step": 2441
    },
    {
      "epoch": 0.1760253730267426,
      "grad_norm": 0.4472295610253784,
      "learning_rate": 0.00018903059072613004,
      "loss": 1.0682,
      "step": 2442
    },
    {
      "epoch": 0.1760974554890795,
      "grad_norm": 0.46974830396678263,
      "learning_rate": 0.0001890199568727746,
      "loss": 1.0994,
      "step": 2443
    },
    {
      "epoch": 0.17616953795141643,
      "grad_norm": 0.11317275746188393,
      "learning_rate": 0.00018900931816703516,
      "loss": 1.1094,
      "step": 2444
    },
    {
      "epoch": 0.17624162041375332,
      "grad_norm": 0.09903411403390228,
      "learning_rate": 0.00018899867460949154,
      "loss": 1.1257,
      "step": 2445
    },
    {
      "epoch": 0.17631370287609024,
      "grad_norm": 0.11670880069518239,
      "learning_rate": 0.00018898802620072398,
      "loss": 1.1955,
      "step": 2446
    },
    {
      "epoch": 0.17638578533842716,
      "grad_norm": 0.6336587281829299,
      "learning_rate": 0.00018897737294131284,
      "loss": 1.0611,
      "step": 2447
    },
    {
      "epoch": 0.17645786780076408,
      "grad_norm": 0.28631974467595117,
      "learning_rate": 0.0001889667148318389,
      "loss": 1.0786,
      "step": 2448
    },
    {
      "epoch": 0.17652995026310098,
      "grad_norm": 1.0799207667174424,
      "learning_rate": 0.00018895605187288305,
      "loss": 1.0957,
      "step": 2449
    },
    {
      "epoch": 0.1766020327254379,
      "grad_norm": 0.12230860577031867,
      "learning_rate": 0.00018894538406502657,
      "loss": 1.1508,
      "step": 2450
    },
    {
      "epoch": 0.17667411518777482,
      "grad_norm": 0.7443935851712135,
      "learning_rate": 0.00018893471140885094,
      "loss": 1.111,
      "step": 2451
    },
    {
      "epoch": 0.17674619765011174,
      "grad_norm": 1.2355272186192574,
      "learning_rate": 0.0001889240339049379,
      "loss": 1.1615,
      "step": 2452
    },
    {
      "epoch": 0.17681828011244863,
      "grad_norm": 0.10082248758373044,
      "learning_rate": 0.00018891335155386946,
      "loss": 1.045,
      "step": 2453
    },
    {
      "epoch": 0.17689036257478555,
      "grad_norm": 0.15859633312111127,
      "learning_rate": 0.00018890266435622792,
      "loss": 1.0588,
      "step": 2454
    },
    {
      "epoch": 0.17696244503712247,
      "grad_norm": 0.11361154565776255,
      "learning_rate": 0.00018889197231259584,
      "loss": 1.103,
      "step": 2455
    },
    {
      "epoch": 0.1770345274994594,
      "grad_norm": 0.36809335693245543,
      "learning_rate": 0.00018888127542355602,
      "loss": 1.0843,
      "step": 2456
    },
    {
      "epoch": 0.1771066099617963,
      "grad_norm": 0.18008455793826209,
      "learning_rate": 0.0001888705736896915,
      "loss": 1.016,
      "step": 2457
    },
    {
      "epoch": 0.1771786924241332,
      "grad_norm": 1.1392753230125547,
      "learning_rate": 0.0001888598671115857,
      "loss": 1.1116,
      "step": 2458
    },
    {
      "epoch": 0.17725077488647012,
      "grad_norm": 0.13438376397811463,
      "learning_rate": 0.00018884915568982217,
      "loss": 1.0831,
      "step": 2459
    },
    {
      "epoch": 0.17732285734880704,
      "grad_norm": 0.17706734931748747,
      "learning_rate": 0.00018883843942498475,
      "loss": 1.0719,
      "step": 2460
    },
    {
      "epoch": 0.17739493981114396,
      "grad_norm": 1.0895357025391352,
      "learning_rate": 0.00018882771831765766,
      "loss": 1.0763,
      "step": 2461
    },
    {
      "epoch": 0.17746702227348085,
      "grad_norm": 1.2128217992383994,
      "learning_rate": 0.00018881699236842522,
      "loss": 1.1287,
      "step": 2462
    },
    {
      "epoch": 0.17753910473581777,
      "grad_norm": 0.20340118290579692,
      "learning_rate": 0.00018880626157787212,
      "loss": 1.0722,
      "step": 2463
    },
    {
      "epoch": 0.1776111871981547,
      "grad_norm": 1.32097917938784,
      "learning_rate": 0.00018879552594658328,
      "loss": 1.2123,
      "step": 2464
    },
    {
      "epoch": 0.1776832696604916,
      "grad_norm": 0.21506142965360944,
      "learning_rate": 0.00018878478547514389,
      "loss": 1.1464,
      "step": 2465
    },
    {
      "epoch": 0.1777553521228285,
      "grad_norm": 0.14557791462974137,
      "learning_rate": 0.0001887740401641394,
      "loss": 1.1394,
      "step": 2466
    },
    {
      "epoch": 0.17782743458516542,
      "grad_norm": 0.13680134685148368,
      "learning_rate": 0.0001887632900141555,
      "loss": 1.131,
      "step": 2467
    },
    {
      "epoch": 0.17789951704750234,
      "grad_norm": 4.378382821313103,
      "learning_rate": 0.00018875253502577825,
      "loss": 1.0402,
      "step": 2468
    },
    {
      "epoch": 0.17797159950983926,
      "grad_norm": 2.51000509416779,
      "learning_rate": 0.0001887417751995938,
      "loss": 1.1881,
      "step": 2469
    },
    {
      "epoch": 0.17804368197217618,
      "grad_norm": 0.22269959003243073,
      "learning_rate": 0.0001887310105361887,
      "loss": 1.1178,
      "step": 2470
    },
    {
      "epoch": 0.17811576443451307,
      "grad_norm": 0.12003827095086164,
      "learning_rate": 0.00018872024103614972,
      "loss": 1.1103,
      "step": 2471
    },
    {
      "epoch": 0.17818784689685,
      "grad_norm": 0.13886430088702728,
      "learning_rate": 0.00018870946670006392,
      "loss": 1.0718,
      "step": 2472
    },
    {
      "epoch": 0.1782599293591869,
      "grad_norm": 0.1512960568260366,
      "learning_rate": 0.00018869868752851852,
      "loss": 1.1197,
      "step": 2473
    },
    {
      "epoch": 0.17833201182152383,
      "grad_norm": 0.3348085347882458,
      "learning_rate": 0.00018868790352210115,
      "loss": 0.9848,
      "step": 2474
    },
    {
      "epoch": 0.17840409428386073,
      "grad_norm": 0.12211047667521892,
      "learning_rate": 0.00018867711468139963,
      "loss": 1.1366,
      "step": 2475
    },
    {
      "epoch": 0.17847617674619765,
      "grad_norm": 0.14563785592525852,
      "learning_rate": 0.00018866632100700197,
      "loss": 1.1533,
      "step": 2476
    },
    {
      "epoch": 0.17854825920853457,
      "grad_norm": 0.23909829643243594,
      "learning_rate": 0.00018865552249949663,
      "loss": 1.1361,
      "step": 2477
    },
    {
      "epoch": 0.17862034167087149,
      "grad_norm": 0.1110890804677812,
      "learning_rate": 0.0001886447191594722,
      "loss": 1.1052,
      "step": 2478
    },
    {
      "epoch": 0.17869242413320838,
      "grad_norm": 0.1480092522671474,
      "learning_rate": 0.00018863391098751747,
      "loss": 1.0416,
      "step": 2479
    },
    {
      "epoch": 0.1787645065955453,
      "grad_norm": 0.6455809477228399,
      "learning_rate": 0.00018862309798422172,
      "loss": 1.1143,
      "step": 2480
    },
    {
      "epoch": 0.17883658905788222,
      "grad_norm": 0.10469872633724972,
      "learning_rate": 0.00018861228015017425,
      "loss": 1.0176,
      "step": 2481
    },
    {
      "epoch": 0.17890867152021914,
      "grad_norm": 0.12217105558371778,
      "learning_rate": 0.00018860145748596477,
      "loss": 1.1115,
      "step": 2482
    },
    {
      "epoch": 0.17898075398255606,
      "grad_norm": 0.19951379075124076,
      "learning_rate": 0.00018859062999218319,
      "loss": 1.1298,
      "step": 2483
    },
    {
      "epoch": 0.17905283644489295,
      "grad_norm": 0.3254776858042462,
      "learning_rate": 0.00018857979766941975,
      "loss": 1.0973,
      "step": 2484
    },
    {
      "epoch": 0.17912491890722987,
      "grad_norm": 0.7209437563918022,
      "learning_rate": 0.00018856896051826487,
      "loss": 1.0884,
      "step": 2485
    },
    {
      "epoch": 0.1791970013695668,
      "grad_norm": 0.17140052849273843,
      "learning_rate": 0.00018855811853930927,
      "loss": 1.004,
      "step": 2486
    },
    {
      "epoch": 0.1792690838319037,
      "grad_norm": 0.12045788336201524,
      "learning_rate": 0.00018854727173314394,
      "loss": 1.0486,
      "step": 2487
    },
    {
      "epoch": 0.1793411662942406,
      "grad_norm": 0.28611179612555176,
      "learning_rate": 0.00018853642010036015,
      "loss": 1.1312,
      "step": 2488
    },
    {
      "epoch": 0.17941324875657752,
      "grad_norm": 0.10999326483728529,
      "learning_rate": 0.0001885255636415494,
      "loss": 1.1155,
      "step": 2489
    },
    {
      "epoch": 0.17948533121891444,
      "grad_norm": 0.14915540608852193,
      "learning_rate": 0.00018851470235730345,
      "loss": 1.0896,
      "step": 2490
    },
    {
      "epoch": 0.17955741368125136,
      "grad_norm": 0.15340335799676605,
      "learning_rate": 0.00018850383624821433,
      "loss": 1.033,
      "step": 2491
    },
    {
      "epoch": 0.17962949614358825,
      "grad_norm": 0.39966742866859306,
      "learning_rate": 0.00018849296531487438,
      "loss": 1.1792,
      "step": 2492
    },
    {
      "epoch": 0.17970157860592517,
      "grad_norm": 0.11362900267844522,
      "learning_rate": 0.00018848208955787613,
      "loss": 1.1614,
      "step": 2493
    },
    {
      "epoch": 0.1797736610682621,
      "grad_norm": 0.3475377703615441,
      "learning_rate": 0.0001884712089778124,
      "loss": 1.1234,
      "step": 2494
    },
    {
      "epoch": 0.179845743530599,
      "grad_norm": 2.4296968698357335,
      "learning_rate": 0.00018846032357527635,
      "loss": 1.1328,
      "step": 2495
    },
    {
      "epoch": 0.17991782599293593,
      "grad_norm": 0.8135289897813954,
      "learning_rate": 0.00018844943335086123,
      "loss": 1.1581,
      "step": 2496
    },
    {
      "epoch": 0.17998990845527282,
      "grad_norm": 0.5256885270884928,
      "learning_rate": 0.0001884385383051607,
      "loss": 1.0876,
      "step": 2497
    },
    {
      "epoch": 0.18006199091760974,
      "grad_norm": 0.11361315900086164,
      "learning_rate": 0.00018842763843876864,
      "loss": 1.0945,
      "step": 2498
    },
    {
      "epoch": 0.18013407337994666,
      "grad_norm": 0.29752978057314305,
      "learning_rate": 0.0001884167337522792,
      "loss": 1.0511,
      "step": 2499
    },
    {
      "epoch": 0.18020615584228358,
      "grad_norm": 0.22372904684259237,
      "learning_rate": 0.00018840582424628678,
      "loss": 1.0838,
      "step": 2500
    },
    {
      "epoch": 0.18027823830462048,
      "grad_norm": 0.15771214977684583,
      "learning_rate": 0.00018839490992138603,
      "loss": 1.083,
      "step": 2501
    },
    {
      "epoch": 0.1803503207669574,
      "grad_norm": 0.7459642363910646,
      "learning_rate": 0.00018838399077817188,
      "loss": 1.0979,
      "step": 2502
    },
    {
      "epoch": 0.18042240322929431,
      "grad_norm": 0.23404795376294396,
      "learning_rate": 0.00018837306681723952,
      "loss": 1.0864,
      "step": 2503
    },
    {
      "epoch": 0.18049448569163123,
      "grad_norm": 0.14446885653023778,
      "learning_rate": 0.00018836213803918444,
      "loss": 1.1069,
      "step": 2504
    },
    {
      "epoch": 0.18056656815396813,
      "grad_norm": 0.3750044736088307,
      "learning_rate": 0.0001883512044446023,
      "loss": 1.0785,
      "step": 2505
    },
    {
      "epoch": 0.18063865061630505,
      "grad_norm": 0.37642612089143124,
      "learning_rate": 0.00018834026603408913,
      "loss": 1.1862,
      "step": 2506
    },
    {
      "epoch": 0.18071073307864197,
      "grad_norm": 0.15415752386272977,
      "learning_rate": 0.00018832932280824112,
      "loss": 1.102,
      "step": 2507
    },
    {
      "epoch": 0.1807828155409789,
      "grad_norm": 0.6778177838408469,
      "learning_rate": 0.00018831837476765482,
      "loss": 1.0517,
      "step": 2508
    },
    {
      "epoch": 0.1808548980033158,
      "grad_norm": 0.23554417425987714,
      "learning_rate": 0.00018830742191292696,
      "loss": 1.0987,
      "step": 2509
    },
    {
      "epoch": 0.1809269804656527,
      "grad_norm": 0.12909575905390608,
      "learning_rate": 0.00018829646424465456,
      "loss": 1.0217,
      "step": 2510
    },
    {
      "epoch": 0.18099906292798962,
      "grad_norm": 0.16336766609091827,
      "learning_rate": 0.00018828550176343498,
      "loss": 1.0686,
      "step": 2511
    },
    {
      "epoch": 0.18107114539032654,
      "grad_norm": 0.1264606830777846,
      "learning_rate": 0.0001882745344698657,
      "loss": 1.128,
      "step": 2512
    },
    {
      "epoch": 0.18114322785266346,
      "grad_norm": 0.16875702506292786,
      "learning_rate": 0.00018826356236454462,
      "loss": 1.0998,
      "step": 2513
    },
    {
      "epoch": 0.18121531031500035,
      "grad_norm": 0.35527464750030346,
      "learning_rate": 0.00018825258544806972,
      "loss": 1.0586,
      "step": 2514
    },
    {
      "epoch": 0.18128739277733727,
      "grad_norm": 0.10709441055828388,
      "learning_rate": 0.00018824160372103935,
      "loss": 1.0993,
      "step": 2515
    },
    {
      "epoch": 0.1813594752396742,
      "grad_norm": 0.43898592347985965,
      "learning_rate": 0.00018823061718405222,
      "loss": 0.9853,
      "step": 2516
    },
    {
      "epoch": 0.1814315577020111,
      "grad_norm": 0.9919581827583012,
      "learning_rate": 0.00018821962583770704,
      "loss": 1.1402,
      "step": 2517
    },
    {
      "epoch": 0.181503640164348,
      "grad_norm": 0.3898310149698375,
      "learning_rate": 0.00018820862968260305,
      "loss": 1.0618,
      "step": 2518
    },
    {
      "epoch": 0.18157572262668492,
      "grad_norm": 0.1128731911921227,
      "learning_rate": 0.00018819762871933967,
      "loss": 1.1233,
      "step": 2519
    },
    {
      "epoch": 0.18164780508902184,
      "grad_norm": 0.22299346137950882,
      "learning_rate": 0.0001881866229485164,
      "loss": 1.1228,
      "step": 2520
    },
    {
      "epoch": 0.18171988755135876,
      "grad_norm": 0.26925985361737553,
      "learning_rate": 0.00018817561237073326,
      "loss": 1.0546,
      "step": 2521
    },
    {
      "epoch": 0.18179197001369568,
      "grad_norm": 0.14284483418895763,
      "learning_rate": 0.00018816459698659045,
      "loss": 1.0385,
      "step": 2522
    },
    {
      "epoch": 0.18186405247603257,
      "grad_norm": 0.1009280223480327,
      "learning_rate": 0.0001881535767966883,
      "loss": 1.17,
      "step": 2523
    },
    {
      "epoch": 0.1819361349383695,
      "grad_norm": 0.5499567654880557,
      "learning_rate": 0.00018814255180162759,
      "loss": 1.0586,
      "step": 2524
    },
    {
      "epoch": 0.1820082174007064,
      "grad_norm": 0.346806872525169,
      "learning_rate": 0.00018813152200200926,
      "loss": 1.1062,
      "step": 2525
    },
    {
      "epoch": 0.18208029986304333,
      "grad_norm": 0.14319677319058977,
      "learning_rate": 0.00018812048739843455,
      "loss": 1.1216,
      "step": 2526
    },
    {
      "epoch": 0.18215238232538022,
      "grad_norm": 0.1106599388320653,
      "learning_rate": 0.00018810944799150492,
      "loss": 1.083,
      "step": 2527
    },
    {
      "epoch": 0.18222446478771714,
      "grad_norm": 0.23190524537497423,
      "learning_rate": 0.00018809840378182212,
      "loss": 1.116,
      "step": 2528
    },
    {
      "epoch": 0.18229654725005406,
      "grad_norm": 0.18293138711264406,
      "learning_rate": 0.00018808735476998818,
      "loss": 1.0721,
      "step": 2529
    },
    {
      "epoch": 0.18236862971239098,
      "grad_norm": 0.842386737497791,
      "learning_rate": 0.00018807630095660533,
      "loss": 1.142,
      "step": 2530
    },
    {
      "epoch": 0.18244071217472788,
      "grad_norm": 0.18207128053117183,
      "learning_rate": 0.00018806524234227614,
      "loss": 1.1025,
      "step": 2531
    },
    {
      "epoch": 0.1825127946370648,
      "grad_norm": 0.11823167346875664,
      "learning_rate": 0.00018805417892760336,
      "loss": 1.0904,
      "step": 2532
    },
    {
      "epoch": 0.18258487709940172,
      "grad_norm": 0.11296514461872997,
      "learning_rate": 0.0001880431107131901,
      "loss": 1.0133,
      "step": 2533
    },
    {
      "epoch": 0.18265695956173864,
      "grad_norm": 0.29715864769672024,
      "learning_rate": 0.00018803203769963967,
      "loss": 1.1704,
      "step": 2534
    },
    {
      "epoch": 0.18272904202407556,
      "grad_norm": 0.20467836673803616,
      "learning_rate": 0.00018802095988755557,
      "loss": 1.0498,
      "step": 2535
    },
    {
      "epoch": 0.18280112448641245,
      "grad_norm": 0.14109672559163852,
      "learning_rate": 0.00018800987727754173,
      "loss": 1.0417,
      "step": 2536
    },
    {
      "epoch": 0.18287320694874937,
      "grad_norm": 0.19313525633257317,
      "learning_rate": 0.00018799878987020224,
      "loss": 1.1752,
      "step": 2537
    },
    {
      "epoch": 0.1829452894110863,
      "grad_norm": 0.13090229756112767,
      "learning_rate": 0.0001879876976661414,
      "loss": 1.1048,
      "step": 2538
    },
    {
      "epoch": 0.1830173718734232,
      "grad_norm": 0.15207925413746493,
      "learning_rate": 0.00018797660066596393,
      "loss": 1.0835,
      "step": 2539
    },
    {
      "epoch": 0.1830894543357601,
      "grad_norm": 0.12480191886165955,
      "learning_rate": 0.0001879654988702746,
      "loss": 1.1634,
      "step": 2540
    },
    {
      "epoch": 0.18316153679809702,
      "grad_norm": 0.11850690436775138,
      "learning_rate": 0.00018795439227967865,
      "loss": 1.1467,
      "step": 2541
    },
    {
      "epoch": 0.18323361926043394,
      "grad_norm": 0.28265178919448475,
      "learning_rate": 0.0001879432808947815,
      "loss": 1.0589,
      "step": 2542
    },
    {
      "epoch": 0.18330570172277086,
      "grad_norm": 0.11037634421855065,
      "learning_rate": 0.00018793216471618874,
      "loss": 1.1213,
      "step": 2543
    },
    {
      "epoch": 0.18337778418510775,
      "grad_norm": 0.18138197030238876,
      "learning_rate": 0.0001879210437445063,
      "loss": 1.1441,
      "step": 2544
    },
    {
      "epoch": 0.18344986664744467,
      "grad_norm": 0.3406343677299468,
      "learning_rate": 0.00018790991798034047,
      "loss": 1.1766,
      "step": 2545
    },
    {
      "epoch": 0.1835219491097816,
      "grad_norm": 0.14238788855583295,
      "learning_rate": 0.00018789878742429762,
      "loss": 1.0518,
      "step": 2546
    },
    {
      "epoch": 0.1835940315721185,
      "grad_norm": 0.15611470006057288,
      "learning_rate": 0.00018788765207698448,
      "loss": 0.9915,
      "step": 2547
    },
    {
      "epoch": 0.18366611403445543,
      "grad_norm": 0.13014080971716419,
      "learning_rate": 0.00018787651193900805,
      "loss": 1.1079,
      "step": 2548
    },
    {
      "epoch": 0.18373819649679232,
      "grad_norm": 1.9678020669254943,
      "learning_rate": 0.00018786536701097557,
      "loss": 1.0807,
      "step": 2549
    },
    {
      "epoch": 0.18381027895912924,
      "grad_norm": 0.10936817315474615,
      "learning_rate": 0.0001878542172934945,
      "loss": 1.0541,
      "step": 2550
    },
    {
      "epoch": 0.18388236142146616,
      "grad_norm": 0.24836978163501738,
      "learning_rate": 0.00018784306278717263,
      "loss": 1.0822,
      "step": 2551
    },
    {
      "epoch": 0.18395444388380308,
      "grad_norm": 0.1215598315898788,
      "learning_rate": 0.00018783190349261797,
      "loss": 1.1416,
      "step": 2552
    },
    {
      "epoch": 0.18402652634613997,
      "grad_norm": 0.2006596314317821,
      "learning_rate": 0.0001878207394104388,
      "loss": 1.0713,
      "step": 2553
    },
    {
      "epoch": 0.1840986088084769,
      "grad_norm": 0.1340524771201856,
      "learning_rate": 0.00018780957054124366,
      "loss": 1.1339,
      "step": 2554
    },
    {
      "epoch": 0.1841706912708138,
      "grad_norm": 0.11714219711863862,
      "learning_rate": 0.00018779839688564138,
      "loss": 1.1139,
      "step": 2555
    },
    {
      "epoch": 0.18424277373315073,
      "grad_norm": 0.16002133125242846,
      "learning_rate": 0.000187787218444241,
      "loss": 1.0952,
      "step": 2556
    },
    {
      "epoch": 0.18431485619548763,
      "grad_norm": 0.2261803568575285,
      "learning_rate": 0.00018777603521765183,
      "loss": 1.1281,
      "step": 2557
    },
    {
      "epoch": 0.18438693865782455,
      "grad_norm": 0.15199011822478536,
      "learning_rate": 0.00018776484720648347,
      "loss": 1.089,
      "step": 2558
    },
    {
      "epoch": 0.18445902112016146,
      "grad_norm": 0.11178731053849797,
      "learning_rate": 0.0001877536544113458,
      "loss": 1.1361,
      "step": 2559
    },
    {
      "epoch": 0.18453110358249838,
      "grad_norm": 0.11247243201573996,
      "learning_rate": 0.00018774245683284889,
      "loss": 1.0785,
      "step": 2560
    },
    {
      "epoch": 0.1846031860448353,
      "grad_norm": 0.11985522511195426,
      "learning_rate": 0.0001877312544716031,
      "loss": 1.1458,
      "step": 2561
    },
    {
      "epoch": 0.1846752685071722,
      "grad_norm": 0.3002478785125748,
      "learning_rate": 0.00018772004732821906,
      "loss": 1.1132,
      "step": 2562
    },
    {
      "epoch": 0.18474735096950912,
      "grad_norm": 0.12429643423034321,
      "learning_rate": 0.0001877088354033077,
      "loss": 1.095,
      "step": 2563
    },
    {
      "epoch": 0.18481943343184604,
      "grad_norm": 0.1666831342106976,
      "learning_rate": 0.00018769761869748015,
      "loss": 1.0696,
      "step": 2564
    },
    {
      "epoch": 0.18489151589418296,
      "grad_norm": 0.1558143685898585,
      "learning_rate": 0.0001876863972113478,
      "loss": 1.0814,
      "step": 2565
    },
    {
      "epoch": 0.18496359835651985,
      "grad_norm": 0.11893972214579997,
      "learning_rate": 0.00018767517094552235,
      "loss": 1.0949,
      "step": 2566
    },
    {
      "epoch": 0.18503568081885677,
      "grad_norm": 0.1554696101831745,
      "learning_rate": 0.0001876639399006157,
      "loss": 1.0972,
      "step": 2567
    },
    {
      "epoch": 0.1851077632811937,
      "grad_norm": 1.256343440666775,
      "learning_rate": 0.0001876527040772401,
      "loss": 1.132,
      "step": 2568
    },
    {
      "epoch": 0.1851798457435306,
      "grad_norm": 0.3178959477568068,
      "learning_rate": 0.00018764146347600793,
      "loss": 1.1688,
      "step": 2569
    },
    {
      "epoch": 0.1852519282058675,
      "grad_norm": 0.11316482173757884,
      "learning_rate": 0.00018763021809753194,
      "loss": 1.0674,
      "step": 2570
    },
    {
      "epoch": 0.18532401066820442,
      "grad_norm": 0.14767695497527972,
      "learning_rate": 0.0001876189679424251,
      "loss": 1.0895,
      "step": 2571
    },
    {
      "epoch": 0.18539609313054134,
      "grad_norm": 0.11449902329799658,
      "learning_rate": 0.00018760771301130067,
      "loss": 1.063,
      "step": 2572
    },
    {
      "epoch": 0.18546817559287826,
      "grad_norm": 0.10108883546749677,
      "learning_rate": 0.00018759645330477208,
      "loss": 1.0035,
      "step": 2573
    },
    {
      "epoch": 0.18554025805521518,
      "grad_norm": 0.6718553052910059,
      "learning_rate": 0.00018758518882345314,
      "loss": 1.1356,
      "step": 2574
    },
    {
      "epoch": 0.18561234051755207,
      "grad_norm": 0.5307913516460244,
      "learning_rate": 0.00018757391956795787,
      "loss": 1.1096,
      "step": 2575
    },
    {
      "epoch": 0.185684422979889,
      "grad_norm": 0.13617924481994378,
      "learning_rate": 0.00018756264553890052,
      "loss": 1.1122,
      "step": 2576
    },
    {
      "epoch": 0.1857565054422259,
      "grad_norm": 0.1950726747560603,
      "learning_rate": 0.00018755136673689562,
      "loss": 1.1719,
      "step": 2577
    },
    {
      "epoch": 0.18582858790456283,
      "grad_norm": 0.32918288235515714,
      "learning_rate": 0.00018754008316255799,
      "loss": 1.113,
      "step": 2578
    },
    {
      "epoch": 0.18590067036689972,
      "grad_norm": 0.150816328977604,
      "learning_rate": 0.00018752879481650264,
      "loss": 1.1831,
      "step": 2579
    },
    {
      "epoch": 0.18597275282923664,
      "grad_norm": 0.11254615943900237,
      "learning_rate": 0.00018751750169934494,
      "loss": 1.0987,
      "step": 2580
    },
    {
      "epoch": 0.18604483529157356,
      "grad_norm": 0.14154793698534937,
      "learning_rate": 0.00018750620381170044,
      "loss": 1.223,
      "step": 2581
    },
    {
      "epoch": 0.18611691775391048,
      "grad_norm": 1.8413749801221118,
      "learning_rate": 0.00018749490115418498,
      "loss": 1.126,
      "step": 2582
    },
    {
      "epoch": 0.18618900021624737,
      "grad_norm": 0.3907574207054216,
      "learning_rate": 0.00018748359372741465,
      "loss": 1.1086,
      "step": 2583
    },
    {
      "epoch": 0.1862610826785843,
      "grad_norm": 0.12154932180397815,
      "learning_rate": 0.00018747228153200582,
      "loss": 1.0899,
      "step": 2584
    },
    {
      "epoch": 0.18633316514092121,
      "grad_norm": 0.20015897073651961,
      "learning_rate": 0.0001874609645685751,
      "loss": 1.1335,
      "step": 2585
    },
    {
      "epoch": 0.18640524760325813,
      "grad_norm": 0.10006650148988001,
      "learning_rate": 0.00018744964283773935,
      "loss": 1.0773,
      "step": 2586
    },
    {
      "epoch": 0.18647733006559505,
      "grad_norm": 0.2042368862141641,
      "learning_rate": 0.00018743831634011573,
      "loss": 0.9929,
      "step": 2587
    },
    {
      "epoch": 0.18654941252793195,
      "grad_norm": 0.25233326804014233,
      "learning_rate": 0.00018742698507632162,
      "loss": 1.0294,
      "step": 2588
    },
    {
      "epoch": 0.18662149499026887,
      "grad_norm": 0.12535013102661513,
      "learning_rate": 0.0001874156490469747,
      "loss": 1.1553,
      "step": 2589
    },
    {
      "epoch": 0.18669357745260579,
      "grad_norm": 0.49253589617940685,
      "learning_rate": 0.0001874043082526928,
      "loss": 1.1306,
      "step": 2590
    },
    {
      "epoch": 0.1867656599149427,
      "grad_norm": 0.3800730813900662,
      "learning_rate": 0.00018739296269409424,
      "loss": 1.0983,
      "step": 2591
    },
    {
      "epoch": 0.1868377423772796,
      "grad_norm": 0.43005980574458436,
      "learning_rate": 0.0001873816123717973,
      "loss": 1.1193,
      "step": 2592
    },
    {
      "epoch": 0.18690982483961652,
      "grad_norm": 0.16368517975119823,
      "learning_rate": 0.00018737025728642078,
      "loss": 1.2228,
      "step": 2593
    },
    {
      "epoch": 0.18698190730195344,
      "grad_norm": 0.2969435817090446,
      "learning_rate": 0.00018735889743858362,
      "loss": 1.0278,
      "step": 2594
    },
    {
      "epoch": 0.18705398976429036,
      "grad_norm": 0.10795202597685837,
      "learning_rate": 0.00018734753282890497,
      "loss": 1.1467,
      "step": 2595
    },
    {
      "epoch": 0.18712607222662725,
      "grad_norm": 0.16630436825840184,
      "learning_rate": 0.0001873361634580044,
      "loss": 1.1876,
      "step": 2596
    },
    {
      "epoch": 0.18719815468896417,
      "grad_norm": 0.10204356192537387,
      "learning_rate": 0.00018732478932650154,
      "loss": 1.0759,
      "step": 2597
    },
    {
      "epoch": 0.1872702371513011,
      "grad_norm": 0.1230751250683547,
      "learning_rate": 0.0001873134104350165,
      "loss": 1.0618,
      "step": 2598
    },
    {
      "epoch": 0.187342319613638,
      "grad_norm": 0.2566395783238416,
      "learning_rate": 0.0001873020267841694,
      "loss": 1.0928,
      "step": 2599
    },
    {
      "epoch": 0.18741440207597493,
      "grad_norm": 0.14919436492516028,
      "learning_rate": 0.00018729063837458082,
      "loss": 1.0296,
      "step": 2600
    },
    {
      "epoch": 0.18748648453831182,
      "grad_norm": 0.11230133412204481,
      "learning_rate": 0.00018727924520687154,
      "loss": 1.1089,
      "step": 2601
    },
    {
      "epoch": 0.18755856700064874,
      "grad_norm": 2.0504119162922665,
      "learning_rate": 0.00018726784728166258,
      "loss": 1.1081,
      "step": 2602
    },
    {
      "epoch": 0.18763064946298566,
      "grad_norm": 0.18012483261667875,
      "learning_rate": 0.0001872564445995752,
      "loss": 1.0357,
      "step": 2603
    },
    {
      "epoch": 0.18770273192532258,
      "grad_norm": 0.13026990943085742,
      "learning_rate": 0.00018724503716123102,
      "loss": 1.1186,
      "step": 2604
    },
    {
      "epoch": 0.18777481438765947,
      "grad_norm": 0.12511507356266702,
      "learning_rate": 0.00018723362496725175,
      "loss": 1.1287,
      "step": 2605
    },
    {
      "epoch": 0.1878468968499964,
      "grad_norm": 0.25345504034175637,
      "learning_rate": 0.0001872222080182595,
      "loss": 1.2214,
      "step": 2606
    },
    {
      "epoch": 0.1879189793123333,
      "grad_norm": 0.12509767511886916,
      "learning_rate": 0.00018721078631487665,
      "loss": 1.1888,
      "step": 2607
    },
    {
      "epoch": 0.18799106177467023,
      "grad_norm": 1.2736546434249878,
      "learning_rate": 0.0001871993598577257,
      "loss": 1.1059,
      "step": 2608
    },
    {
      "epoch": 0.18806314423700712,
      "grad_norm": 0.13544851852461604,
      "learning_rate": 0.00018718792864742953,
      "loss": 1.1097,
      "step": 2609
    },
    {
      "epoch": 0.18813522669934404,
      "grad_norm": 1.5759993858716277,
      "learning_rate": 0.00018717649268461126,
      "loss": 1.0352,
      "step": 2610
    },
    {
      "epoch": 0.18820730916168096,
      "grad_norm": 0.12508745953401448,
      "learning_rate": 0.0001871650519698942,
      "loss": 1.1606,
      "step": 2611
    },
    {
      "epoch": 0.18827939162401788,
      "grad_norm": 0.10572300888148646,
      "learning_rate": 0.00018715360650390205,
      "loss": 1.017,
      "step": 2612
    },
    {
      "epoch": 0.1883514740863548,
      "grad_norm": 0.3871157003190603,
      "learning_rate": 0.0001871421562872586,
      "loss": 1.0458,
      "step": 2613
    },
    {
      "epoch": 0.1884235565486917,
      "grad_norm": 0.10740244720927039,
      "learning_rate": 0.00018713070132058806,
      "loss": 1.1149,
      "step": 2614
    },
    {
      "epoch": 0.18849563901102862,
      "grad_norm": 0.21803691743727943,
      "learning_rate": 0.00018711924160451485,
      "loss": 1.1844,
      "step": 2615
    },
    {
      "epoch": 0.18856772147336553,
      "grad_norm": 0.14750243990745165,
      "learning_rate": 0.00018710777713966352,
      "loss": 1.0524,
      "step": 2616
    },
    {
      "epoch": 0.18863980393570245,
      "grad_norm": 0.1120314757167524,
      "learning_rate": 0.00018709630792665906,
      "loss": 1.0455,
      "step": 2617
    },
    {
      "epoch": 0.18871188639803935,
      "grad_norm": 0.12287005539074133,
      "learning_rate": 0.00018708483396612664,
      "loss": 1.116,
      "step": 2618
    },
    {
      "epoch": 0.18878396886037627,
      "grad_norm": 0.11632502142918429,
      "learning_rate": 0.0001870733552586917,
      "loss": 1.1236,
      "step": 2619
    },
    {
      "epoch": 0.1888560513227132,
      "grad_norm": 0.3623203794846796,
      "learning_rate": 0.0001870618718049799,
      "loss": 1.1691,
      "step": 2620
    },
    {
      "epoch": 0.1889281337850501,
      "grad_norm": 0.13093310807230726,
      "learning_rate": 0.0001870503836056172,
      "loss": 1.0488,
      "step": 2621
    },
    {
      "epoch": 0.189000216247387,
      "grad_norm": 0.19625308434680386,
      "learning_rate": 0.00018703889066122985,
      "loss": 1.1473,
      "step": 2622
    },
    {
      "epoch": 0.18907229870972392,
      "grad_norm": 0.15803797043560658,
      "learning_rate": 0.00018702739297244425,
      "loss": 1.0652,
      "step": 2623
    },
    {
      "epoch": 0.18914438117206084,
      "grad_norm": 0.23105433936163147,
      "learning_rate": 0.0001870158905398872,
      "loss": 1.1448,
      "step": 2624
    },
    {
      "epoch": 0.18921646363439776,
      "grad_norm": 0.23526390528677812,
      "learning_rate": 0.0001870043833641856,
      "loss": 1.0937,
      "step": 2625
    },
    {
      "epoch": 0.18928854609673468,
      "grad_norm": 0.14271707045339802,
      "learning_rate": 0.0001869928714459668,
      "loss": 1.1001,
      "step": 2626
    },
    {
      "epoch": 0.18936062855907157,
      "grad_norm": 0.39369893990374993,
      "learning_rate": 0.00018698135478585823,
      "loss": 1.1438,
      "step": 2627
    },
    {
      "epoch": 0.1894327110214085,
      "grad_norm": 0.1665372902278685,
      "learning_rate": 0.00018696983338448763,
      "loss": 1.0785,
      "step": 2628
    },
    {
      "epoch": 0.1895047934837454,
      "grad_norm": 0.11187187057471444,
      "learning_rate": 0.00018695830724248312,
      "loss": 1.0519,
      "step": 2629
    },
    {
      "epoch": 0.18957687594608233,
      "grad_norm": 0.12973878992267707,
      "learning_rate": 0.00018694677636047285,
      "loss": 1.1516,
      "step": 2630
    },
    {
      "epoch": 0.18964895840841922,
      "grad_norm": 0.14135531668574766,
      "learning_rate": 0.00018693524073908546,
      "loss": 1.0344,
      "step": 2631
    },
    {
      "epoch": 0.18972104087075614,
      "grad_norm": 0.1367246235399504,
      "learning_rate": 0.0001869237003789497,
      "loss": 1.0319,
      "step": 2632
    },
    {
      "epoch": 0.18979312333309306,
      "grad_norm": 0.3759505984117836,
      "learning_rate": 0.0001869121552806946,
      "loss": 1.0164,
      "step": 2633
    },
    {
      "epoch": 0.18986520579542998,
      "grad_norm": 0.16217114397746374,
      "learning_rate": 0.00018690060544494954,
      "loss": 1.0035,
      "step": 2634
    },
    {
      "epoch": 0.18993728825776687,
      "grad_norm": 0.12300908800876448,
      "learning_rate": 0.000186889050872344,
      "loss": 1.1438,
      "step": 2635
    },
    {
      "epoch": 0.1900093707201038,
      "grad_norm": 0.1421125817437004,
      "learning_rate": 0.0001868774915635079,
      "loss": 1.0677,
      "step": 2636
    },
    {
      "epoch": 0.1900814531824407,
      "grad_norm": 0.1259247303145183,
      "learning_rate": 0.00018686592751907127,
      "loss": 1.0284,
      "step": 2637
    },
    {
      "epoch": 0.19015353564477763,
      "grad_norm": 0.12309558861594731,
      "learning_rate": 0.00018685435873966446,
      "loss": 1.1272,
      "step": 2638
    },
    {
      "epoch": 0.19022561810711455,
      "grad_norm": 0.5044206726444658,
      "learning_rate": 0.00018684278522591806,
      "loss": 1.0292,
      "step": 2639
    },
    {
      "epoch": 0.19029770056945144,
      "grad_norm": 0.18046540589291352,
      "learning_rate": 0.00018683120697846297,
      "loss": 1.0538,
      "step": 2640
    },
    {
      "epoch": 0.19036978303178836,
      "grad_norm": 0.17116398335451755,
      "learning_rate": 0.0001868196239979303,
      "loss": 1.1038,
      "step": 2641
    },
    {
      "epoch": 0.19044186549412528,
      "grad_norm": 0.24556883649428748,
      "learning_rate": 0.00018680803628495137,
      "loss": 1.1039,
      "step": 2642
    },
    {
      "epoch": 0.1905139479564622,
      "grad_norm": 0.1181386223816648,
      "learning_rate": 0.00018679644384015788,
      "loss": 1.2016,
      "step": 2643
    },
    {
      "epoch": 0.1905860304187991,
      "grad_norm": 1.389722223517908,
      "learning_rate": 0.00018678484666418167,
      "loss": 1.1353,
      "step": 2644
    },
    {
      "epoch": 0.19065811288113602,
      "grad_norm": 0.14735605521130285,
      "learning_rate": 0.00018677324475765498,
      "loss": 1.066,
      "step": 2645
    },
    {
      "epoch": 0.19073019534347294,
      "grad_norm": 0.11732192623954325,
      "learning_rate": 0.0001867616381212101,
      "loss": 1.0166,
      "step": 2646
    },
    {
      "epoch": 0.19080227780580986,
      "grad_norm": 0.6837513895916927,
      "learning_rate": 0.00018675002675547976,
      "loss": 1.1666,
      "step": 2647
    },
    {
      "epoch": 0.19087436026814675,
      "grad_norm": 0.18787026666326945,
      "learning_rate": 0.00018673841066109688,
      "loss": 1.0673,
      "step": 2648
    },
    {
      "epoch": 0.19094644273048367,
      "grad_norm": 0.2865918420836228,
      "learning_rate": 0.00018672678983869463,
      "loss": 1.14,
      "step": 2649
    },
    {
      "epoch": 0.1910185251928206,
      "grad_norm": 0.1484790281413326,
      "learning_rate": 0.00018671516428890648,
      "loss": 1.1422,
      "step": 2650
    },
    {
      "epoch": 0.1910906076551575,
      "grad_norm": 0.10805776268557321,
      "learning_rate": 0.0001867035340123661,
      "loss": 1.1034,
      "step": 2651
    },
    {
      "epoch": 0.19116269011749443,
      "grad_norm": 0.13427745578445927,
      "learning_rate": 0.0001866918990097074,
      "loss": 1.0962,
      "step": 2652
    },
    {
      "epoch": 0.19123477257983132,
      "grad_norm": 0.11473281736941204,
      "learning_rate": 0.00018668025928156466,
      "loss": 1.0832,
      "step": 2653
    },
    {
      "epoch": 0.19130685504216824,
      "grad_norm": 0.13036390193688457,
      "learning_rate": 0.00018666861482857235,
      "loss": 1.0666,
      "step": 2654
    },
    {
      "epoch": 0.19137893750450516,
      "grad_norm": 0.12405923456564605,
      "learning_rate": 0.00018665696565136513,
      "loss": 1.0685,
      "step": 2655
    },
    {
      "epoch": 0.19145101996684208,
      "grad_norm": 0.11576378256413981,
      "learning_rate": 0.00018664531175057803,
      "loss": 1.1512,
      "step": 2656
    },
    {
      "epoch": 0.19152310242917897,
      "grad_norm": 0.17495034883317778,
      "learning_rate": 0.00018663365312684632,
      "loss": 1.1243,
      "step": 2657
    },
    {
      "epoch": 0.1915951848915159,
      "grad_norm": 0.26621146312272137,
      "learning_rate": 0.00018662198978080543,
      "loss": 1.1874,
      "step": 2658
    },
    {
      "epoch": 0.1916672673538528,
      "grad_norm": 0.24553843957922744,
      "learning_rate": 0.00018661032171309118,
      "loss": 1.1003,
      "step": 2659
    },
    {
      "epoch": 0.19173934981618973,
      "grad_norm": 0.6438036187519516,
      "learning_rate": 0.00018659864892433956,
      "loss": 1.1298,
      "step": 2660
    },
    {
      "epoch": 0.19181143227852662,
      "grad_norm": 0.15856322501916612,
      "learning_rate": 0.00018658697141518685,
      "loss": 1.087,
      "step": 2661
    },
    {
      "epoch": 0.19188351474086354,
      "grad_norm": 0.28969121010729876,
      "learning_rate": 0.00018657528918626954,
      "loss": 1.1312,
      "step": 2662
    },
    {
      "epoch": 0.19195559720320046,
      "grad_norm": 0.11908328323970051,
      "learning_rate": 0.00018656360223822445,
      "loss": 1.1072,
      "step": 2663
    },
    {
      "epoch": 0.19202767966553738,
      "grad_norm": 0.21155851219803334,
      "learning_rate": 0.00018655191057168863,
      "loss": 1.0572,
      "step": 2664
    },
    {
      "epoch": 0.1920997621278743,
      "grad_norm": 0.11645748845121894,
      "learning_rate": 0.00018654021418729937,
      "loss": 1.127,
      "step": 2665
    },
    {
      "epoch": 0.1921718445902112,
      "grad_norm": 0.11457326999994387,
      "learning_rate": 0.00018652851308569423,
      "loss": 1.2276,
      "step": 2666
    },
    {
      "epoch": 0.1922439270525481,
      "grad_norm": 0.1282603310835487,
      "learning_rate": 0.000186516807267511,
      "loss": 1.0993,
      "step": 2667
    },
    {
      "epoch": 0.19231600951488503,
      "grad_norm": 0.11593094607459507,
      "learning_rate": 0.0001865050967333878,
      "loss": 1.0944,
      "step": 2668
    },
    {
      "epoch": 0.19238809197722195,
      "grad_norm": 0.1098750613035653,
      "learning_rate": 0.00018649338148396289,
      "loss": 1.0679,
      "step": 2669
    },
    {
      "epoch": 0.19246017443955885,
      "grad_norm": 0.2164759552430772,
      "learning_rate": 0.0001864816615198749,
      "loss": 1.1384,
      "step": 2670
    },
    {
      "epoch": 0.19253225690189577,
      "grad_norm": 1.6331962329430703,
      "learning_rate": 0.0001864699368417627,
      "loss": 1.0918,
      "step": 2671
    },
    {
      "epoch": 0.19260433936423269,
      "grad_norm": 0.13210250160162706,
      "learning_rate": 0.00018645820745026538,
      "loss": 1.0606,
      "step": 2672
    },
    {
      "epoch": 0.1926764218265696,
      "grad_norm": 0.1463461907204039,
      "learning_rate": 0.00018644647334602223,
      "loss": 1.1216,
      "step": 2673
    },
    {
      "epoch": 0.1927485042889065,
      "grad_norm": 0.15245479123839342,
      "learning_rate": 0.00018643473452967295,
      "loss": 1.1107,
      "step": 2674
    },
    {
      "epoch": 0.19282058675124342,
      "grad_norm": 0.11300222223206584,
      "learning_rate": 0.00018642299100185734,
      "loss": 1.1083,
      "step": 2675
    },
    {
      "epoch": 0.19289266921358034,
      "grad_norm": 0.2062854696623696,
      "learning_rate": 0.00018641124276321555,
      "loss": 1.1274,
      "step": 2676
    },
    {
      "epoch": 0.19296475167591726,
      "grad_norm": 0.22097043497346203,
      "learning_rate": 0.00018639948981438799,
      "loss": 1.1367,
      "step": 2677
    },
    {
      "epoch": 0.19303683413825418,
      "grad_norm": 0.17304201825421273,
      "learning_rate": 0.00018638773215601527,
      "loss": 1.1278,
      "step": 2678
    },
    {
      "epoch": 0.19310891660059107,
      "grad_norm": 0.13979389964569341,
      "learning_rate": 0.00018637596978873835,
      "loss": 1.0856,
      "step": 2679
    },
    {
      "epoch": 0.193180999062928,
      "grad_norm": 0.12616682282520847,
      "learning_rate": 0.0001863642027131983,
      "loss": 1.0471,
      "step": 2680
    },
    {
      "epoch": 0.1932530815252649,
      "grad_norm": 0.12107201839646343,
      "learning_rate": 0.00018635243093003654,
      "loss": 1.1012,
      "step": 2681
    },
    {
      "epoch": 0.19332516398760183,
      "grad_norm": 1.0335455426609972,
      "learning_rate": 0.0001863406544398948,
      "loss": 1.0862,
      "step": 2682
    },
    {
      "epoch": 0.19339724644993872,
      "grad_norm": 0.13029917691605442,
      "learning_rate": 0.00018632887324341498,
      "loss": 1.0675,
      "step": 2683
    },
    {
      "epoch": 0.19346932891227564,
      "grad_norm": 0.12735472305054923,
      "learning_rate": 0.0001863170873412392,
      "loss": 1.1602,
      "step": 2684
    },
    {
      "epoch": 0.19354141137461256,
      "grad_norm": 0.27076528688975193,
      "learning_rate": 0.00018630529673400998,
      "loss": 1.0239,
      "step": 2685
    },
    {
      "epoch": 0.19361349383694948,
      "grad_norm": 0.3398550933706083,
      "learning_rate": 0.00018629350142236993,
      "loss": 1.0376,
      "step": 2686
    },
    {
      "epoch": 0.19368557629928637,
      "grad_norm": 0.17913937229036767,
      "learning_rate": 0.00018628170140696213,
      "loss": 1.0947,
      "step": 2687
    },
    {
      "epoch": 0.1937576587616233,
      "grad_norm": 0.17851498988573558,
      "learning_rate": 0.00018626989668842965,
      "loss": 1.0807,
      "step": 2688
    },
    {
      "epoch": 0.1938297412239602,
      "grad_norm": 0.14482628253232527,
      "learning_rate": 0.000186258087267416,
      "loss": 1.1011,
      "step": 2689
    },
    {
      "epoch": 0.19390182368629713,
      "grad_norm": 0.14477336521142442,
      "learning_rate": 0.00018624627314456493,
      "loss": 1.0363,
      "step": 2690
    },
    {
      "epoch": 0.19397390614863405,
      "grad_norm": 0.12772093591178277,
      "learning_rate": 0.0001862344543205204,
      "loss": 1.0787,
      "step": 2691
    },
    {
      "epoch": 0.19404598861097094,
      "grad_norm": 0.12902187311979324,
      "learning_rate": 0.00018622263079592664,
      "loss": 1.0663,
      "step": 2692
    },
    {
      "epoch": 0.19411807107330786,
      "grad_norm": 0.4415201786815838,
      "learning_rate": 0.00018621080257142812,
      "loss": 1.1062,
      "step": 2693
    },
    {
      "epoch": 0.19419015353564478,
      "grad_norm": 0.14829884038961422,
      "learning_rate": 0.00018619896964766955,
      "loss": 1.1458,
      "step": 2694
    },
    {
      "epoch": 0.1942622359979817,
      "grad_norm": 0.13254233499925827,
      "learning_rate": 0.00018618713202529606,
      "loss": 1.0969,
      "step": 2695
    },
    {
      "epoch": 0.1943343184603186,
      "grad_norm": 0.2079499170442984,
      "learning_rate": 0.00018617528970495278,
      "loss": 1.0707,
      "step": 2696
    },
    {
      "epoch": 0.19440640092265551,
      "grad_norm": 0.10899332352526202,
      "learning_rate": 0.00018616344268728523,
      "loss": 1.1207,
      "step": 2697
    },
    {
      "epoch": 0.19447848338499243,
      "grad_norm": 0.1287756393876003,
      "learning_rate": 0.00018615159097293927,
      "loss": 1.0853,
      "step": 2698
    },
    {
      "epoch": 0.19455056584732935,
      "grad_norm": 0.15426741999407215,
      "learning_rate": 0.00018613973456256083,
      "loss": 1.0158,
      "step": 2699
    },
    {
      "epoch": 0.19462264830966625,
      "grad_norm": 0.14446484893671835,
      "learning_rate": 0.00018612787345679625,
      "loss": 1.0947,
      "step": 2700
    },
    {
      "epoch": 0.19469473077200317,
      "grad_norm": 0.12117078182116955,
      "learning_rate": 0.00018611600765629205,
      "loss": 1.1295,
      "step": 2701
    },
    {
      "epoch": 0.19476681323434009,
      "grad_norm": 0.2181337918727651,
      "learning_rate": 0.00018610413716169499,
      "loss": 1.0853,
      "step": 2702
    },
    {
      "epoch": 0.194838895696677,
      "grad_norm": 0.4447654023479629,
      "learning_rate": 0.00018609226197365217,
      "loss": 1.1956,
      "step": 2703
    },
    {
      "epoch": 0.19491097815901393,
      "grad_norm": 0.28138891168315633,
      "learning_rate": 0.00018608038209281085,
      "loss": 1.0772,
      "step": 2704
    },
    {
      "epoch": 0.19498306062135082,
      "grad_norm": 0.38882791567671077,
      "learning_rate": 0.00018606849751981862,
      "loss": 1.0759,
      "step": 2705
    },
    {
      "epoch": 0.19505514308368774,
      "grad_norm": 0.12396288325810026,
      "learning_rate": 0.0001860566082553233,
      "loss": 1.1027,
      "step": 2706
    },
    {
      "epoch": 0.19512722554602466,
      "grad_norm": 0.680792681730099,
      "learning_rate": 0.0001860447142999729,
      "loss": 1.1378,
      "step": 2707
    },
    {
      "epoch": 0.19519930800836158,
      "grad_norm": 0.1664037584439144,
      "learning_rate": 0.00018603281565441585,
      "loss": 1.1409,
      "step": 2708
    },
    {
      "epoch": 0.19527139047069847,
      "grad_norm": 0.1389186288624432,
      "learning_rate": 0.00018602091231930063,
      "loss": 0.9769,
      "step": 2709
    },
    {
      "epoch": 0.1953434729330354,
      "grad_norm": 0.13348633693101938,
      "learning_rate": 0.00018600900429527616,
      "loss": 1.1421,
      "step": 2710
    },
    {
      "epoch": 0.1954155553953723,
      "grad_norm": 0.26655965141149046,
      "learning_rate": 0.00018599709158299146,
      "loss": 1.0823,
      "step": 2711
    },
    {
      "epoch": 0.19548763785770923,
      "grad_norm": 0.1404829341036754,
      "learning_rate": 0.00018598517418309593,
      "loss": 1.025,
      "step": 2712
    },
    {
      "epoch": 0.19555972032004612,
      "grad_norm": 0.23714082810894682,
      "learning_rate": 0.00018597325209623918,
      "loss": 1.1281,
      "step": 2713
    },
    {
      "epoch": 0.19563180278238304,
      "grad_norm": 0.15566040577010512,
      "learning_rate": 0.00018596132532307103,
      "loss": 1.1107,
      "step": 2714
    },
    {
      "epoch": 0.19570388524471996,
      "grad_norm": 0.11635278079821301,
      "learning_rate": 0.00018594939386424163,
      "loss": 1.029,
      "step": 2715
    },
    {
      "epoch": 0.19577596770705688,
      "grad_norm": 0.11003858959203328,
      "learning_rate": 0.00018593745772040135,
      "loss": 1.0785,
      "step": 2716
    },
    {
      "epoch": 0.1958480501693938,
      "grad_norm": 0.14176879117408375,
      "learning_rate": 0.00018592551689220076,
      "loss": 1.114,
      "step": 2717
    },
    {
      "epoch": 0.1959201326317307,
      "grad_norm": 0.09975910480037146,
      "learning_rate": 0.00018591357138029083,
      "loss": 1.0637,
      "step": 2718
    },
    {
      "epoch": 0.1959922150940676,
      "grad_norm": 1.201490160351798,
      "learning_rate": 0.00018590162118532267,
      "loss": 1.1177,
      "step": 2719
    },
    {
      "epoch": 0.19606429755640453,
      "grad_norm": 0.24825261804914334,
      "learning_rate": 0.0001858896663079476,
      "loss": 1.1259,
      "step": 2720
    },
    {
      "epoch": 0.19613638001874145,
      "grad_norm": 0.28395404704522914,
      "learning_rate": 0.00018587770674881738,
      "loss": 1.0811,
      "step": 2721
    },
    {
      "epoch": 0.19620846248107834,
      "grad_norm": 0.14762939171068984,
      "learning_rate": 0.00018586574250858382,
      "loss": 1.0229,
      "step": 2722
    },
    {
      "epoch": 0.19628054494341526,
      "grad_norm": 0.16873024151127858,
      "learning_rate": 0.00018585377358789913,
      "loss": 1.0769,
      "step": 2723
    },
    {
      "epoch": 0.19635262740575218,
      "grad_norm": 0.11907854242072606,
      "learning_rate": 0.0001858417999874157,
      "loss": 1.1465,
      "step": 2724
    },
    {
      "epoch": 0.1964247098680891,
      "grad_norm": 1.9856592957123569,
      "learning_rate": 0.00018582982170778622,
      "loss": 1.1014,
      "step": 2725
    },
    {
      "epoch": 0.196496792330426,
      "grad_norm": 0.21938773058164449,
      "learning_rate": 0.00018581783874966362,
      "loss": 1.2,
      "step": 2726
    },
    {
      "epoch": 0.19656887479276292,
      "grad_norm": 0.12473898065713084,
      "learning_rate": 0.00018580585111370104,
      "loss": 1.0731,
      "step": 2727
    },
    {
      "epoch": 0.19664095725509984,
      "grad_norm": 1.0747386807795338,
      "learning_rate": 0.00018579385880055193,
      "loss": 1.1016,
      "step": 2728
    },
    {
      "epoch": 0.19671303971743676,
      "grad_norm": 0.1241272893216723,
      "learning_rate": 0.00018578186181086997,
      "loss": 1.0495,
      "step": 2729
    },
    {
      "epoch": 0.19678512217977367,
      "grad_norm": 0.17698943809928838,
      "learning_rate": 0.00018576986014530916,
      "loss": 1.1179,
      "step": 2730
    },
    {
      "epoch": 0.19685720464211057,
      "grad_norm": 0.15108402890323197,
      "learning_rate": 0.0001857578538045236,
      "loss": 1.106,
      "step": 2731
    },
    {
      "epoch": 0.1969292871044475,
      "grad_norm": 0.14128785603242175,
      "learning_rate": 0.0001857458427891678,
      "loss": 1.0966,
      "step": 2732
    },
    {
      "epoch": 0.1970013695667844,
      "grad_norm": 0.23087868968226302,
      "learning_rate": 0.0001857338270998965,
      "loss": 1.1441,
      "step": 2733
    },
    {
      "epoch": 0.19707345202912133,
      "grad_norm": 0.21364087521365185,
      "learning_rate": 0.0001857218067373646,
      "loss": 1.1473,
      "step": 2734
    },
    {
      "epoch": 0.19714553449145822,
      "grad_norm": 0.18453898457302784,
      "learning_rate": 0.00018570978170222736,
      "loss": 1.1247,
      "step": 2735
    },
    {
      "epoch": 0.19721761695379514,
      "grad_norm": 3.8803318752802287,
      "learning_rate": 0.00018569775199514022,
      "loss": 1.128,
      "step": 2736
    },
    {
      "epoch": 0.19728969941613206,
      "grad_norm": 0.1219448346390306,
      "learning_rate": 0.00018568571761675893,
      "loss": 1.0202,
      "step": 2737
    },
    {
      "epoch": 0.19736178187846898,
      "grad_norm": 0.1965595360634272,
      "learning_rate": 0.00018567367856773948,
      "loss": 1.1329,
      "step": 2738
    },
    {
      "epoch": 0.19743386434080587,
      "grad_norm": 0.23709180796396923,
      "learning_rate": 0.00018566163484873806,
      "loss": 0.9883,
      "step": 2739
    },
    {
      "epoch": 0.1975059468031428,
      "grad_norm": 0.20859708611865782,
      "learning_rate": 0.0001856495864604112,
      "loss": 1.0959,
      "step": 2740
    },
    {
      "epoch": 0.1975780292654797,
      "grad_norm": 0.4289433126214025,
      "learning_rate": 0.00018563753340341562,
      "loss": 1.0636,
      "step": 2741
    },
    {
      "epoch": 0.19765011172781663,
      "grad_norm": 0.16342081742060696,
      "learning_rate": 0.0001856254756784084,
      "loss": 1.1164,
      "step": 2742
    },
    {
      "epoch": 0.19772219419015352,
      "grad_norm": 0.949148451019074,
      "learning_rate": 0.00018561341328604666,
      "loss": 1.1466,
      "step": 2743
    },
    {
      "epoch": 0.19779427665249044,
      "grad_norm": 1.9940354989751041,
      "learning_rate": 0.000185601346226988,
      "loss": 1.1295,
      "step": 2744
    },
    {
      "epoch": 0.19786635911482736,
      "grad_norm": 0.13834422402389526,
      "learning_rate": 0.00018558927450189013,
      "loss": 1.1374,
      "step": 2745
    },
    {
      "epoch": 0.19793844157716428,
      "grad_norm": 0.23523103456950595,
      "learning_rate": 0.00018557719811141115,
      "loss": 1.0547,
      "step": 2746
    },
    {
      "epoch": 0.1980105240395012,
      "grad_norm": 0.3663472515453584,
      "learning_rate": 0.00018556511705620924,
      "loss": 0.9735,
      "step": 2747
    },
    {
      "epoch": 0.1980826065018381,
      "grad_norm": 0.671734187535592,
      "learning_rate": 0.00018555303133694297,
      "loss": 1.0834,
      "step": 2748
    },
    {
      "epoch": 0.198154688964175,
      "grad_norm": 3.666560872256196,
      "learning_rate": 0.0001855409409542711,
      "loss": 1.1454,
      "step": 2749
    },
    {
      "epoch": 0.19822677142651193,
      "grad_norm": 0.16299954993238158,
      "learning_rate": 0.00018552884590885272,
      "loss": 1.0362,
      "step": 2750
    },
    {
      "epoch": 0.19829885388884885,
      "grad_norm": 0.11592209745216854,
      "learning_rate": 0.00018551674620134704,
      "loss": 1.0625,
      "step": 2751
    },
    {
      "epoch": 0.19837093635118574,
      "grad_norm": 0.36522370303144486,
      "learning_rate": 0.00018550464183241365,
      "loss": 1.1152,
      "step": 2752
    },
    {
      "epoch": 0.19844301881352266,
      "grad_norm": 1.5036152085870071,
      "learning_rate": 0.00018549253280271231,
      "loss": 1.1771,
      "step": 2753
    },
    {
      "epoch": 0.19851510127585958,
      "grad_norm": 0.2708050167672266,
      "learning_rate": 0.0001854804191129031,
      "loss": 1.0309,
      "step": 2754
    },
    {
      "epoch": 0.1985871837381965,
      "grad_norm": 0.21767675394980887,
      "learning_rate": 0.00018546830076364635,
      "loss": 1.0468,
      "step": 2755
    },
    {
      "epoch": 0.1986592662005334,
      "grad_norm": 0.29638326239727586,
      "learning_rate": 0.00018545617775560258,
      "loss": 1.0646,
      "step": 2756
    },
    {
      "epoch": 0.19873134866287032,
      "grad_norm": 0.27734453956835803,
      "learning_rate": 0.0001854440500894326,
      "loss": 1.1047,
      "step": 2757
    },
    {
      "epoch": 0.19880343112520724,
      "grad_norm": 2.0073865058386797,
      "learning_rate": 0.00018543191776579745,
      "loss": 1.0202,
      "step": 2758
    },
    {
      "epoch": 0.19887551358754416,
      "grad_norm": 0.6417340533329223,
      "learning_rate": 0.00018541978078535852,
      "loss": 1.1366,
      "step": 2759
    },
    {
      "epoch": 0.19894759604988108,
      "grad_norm": 0.9629528609439844,
      "learning_rate": 0.00018540763914877737,
      "loss": 1.0989,
      "step": 2760
    },
    {
      "epoch": 0.19901967851221797,
      "grad_norm": 2.123381155985725,
      "learning_rate": 0.0001853954928567158,
      "loss": 1.1462,
      "step": 2761
    },
    {
      "epoch": 0.1990917609745549,
      "grad_norm": 0.16190355955445784,
      "learning_rate": 0.00018538334190983585,
      "loss": 1.0789,
      "step": 2762
    },
    {
      "epoch": 0.1991638434368918,
      "grad_norm": 0.9376952762957536,
      "learning_rate": 0.00018537118630879994,
      "loss": 1.0799,
      "step": 2763
    },
    {
      "epoch": 0.19923592589922873,
      "grad_norm": 0.3505107836593388,
      "learning_rate": 0.00018535902605427065,
      "loss": 1.0704,
      "step": 2764
    },
    {
      "epoch": 0.19930800836156562,
      "grad_norm": 0.3001707637358861,
      "learning_rate": 0.00018534686114691078,
      "loss": 1.103,
      "step": 2765
    },
    {
      "epoch": 0.19938009082390254,
      "grad_norm": 0.7653036780509103,
      "learning_rate": 0.00018533469158738344,
      "loss": 1.125,
      "step": 2766
    },
    {
      "epoch": 0.19945217328623946,
      "grad_norm": 0.3223080715557558,
      "learning_rate": 0.00018532251737635196,
      "loss": 1.0719,
      "step": 2767
    },
    {
      "epoch": 0.19952425574857638,
      "grad_norm": 0.7920121065913631,
      "learning_rate": 0.00018531033851447996,
      "loss": 1.1929,
      "step": 2768
    },
    {
      "epoch": 0.19959633821091327,
      "grad_norm": 1.9833909512967565,
      "learning_rate": 0.00018529815500243135,
      "loss": 1.04,
      "step": 2769
    },
    {
      "epoch": 0.1996684206732502,
      "grad_norm": 0.20452157787680914,
      "learning_rate": 0.00018528596684087015,
      "loss": 1.0079,
      "step": 2770
    },
    {
      "epoch": 0.1997405031355871,
      "grad_norm": 0.4873492965997476,
      "learning_rate": 0.0001852737740304608,
      "loss": 1.1237,
      "step": 2771
    },
    {
      "epoch": 0.19981258559792403,
      "grad_norm": 0.23925273841461542,
      "learning_rate": 0.00018526157657186787,
      "loss": 1.0891,
      "step": 2772
    },
    {
      "epoch": 0.19988466806026095,
      "grad_norm": 4.07396508277247,
      "learning_rate": 0.00018524937446575624,
      "loss": 1.0272,
      "step": 2773
    },
    {
      "epoch": 0.19995675052259784,
      "grad_norm": 0.11805688927662884,
      "learning_rate": 0.00018523716771279104,
      "loss": 1.0834,
      "step": 2774
    },
    {
      "epoch": 0.20002883298493476,
      "grad_norm": 0.5399277291172505,
      "learning_rate": 0.00018522495631363768,
      "loss": 1.0699,
      "step": 2775
    },
    {
      "epoch": 0.20010091544727168,
      "grad_norm": 0.1602043000354703,
      "learning_rate": 0.00018521274026896172,
      "loss": 1.1426,
      "step": 2776
    },
    {
      "epoch": 0.2001729979096086,
      "grad_norm": 0.2068563780023728,
      "learning_rate": 0.00018520051957942907,
      "loss": 1.1082,
      "step": 2777
    },
    {
      "epoch": 0.2002450803719455,
      "grad_norm": 2.7568874329802635,
      "learning_rate": 0.0001851882942457059,
      "loss": 1.0878,
      "step": 2778
    },
    {
      "epoch": 0.20031716283428241,
      "grad_norm": 0.41042660381560636,
      "learning_rate": 0.00018517606426845856,
      "loss": 1.1668,
      "step": 2779
    },
    {
      "epoch": 0.20038924529661933,
      "grad_norm": 0.9051317551006569,
      "learning_rate": 0.00018516382964835372,
      "loss": 1.1471,
      "step": 2780
    },
    {
      "epoch": 0.20046132775895625,
      "grad_norm": 2.690563751724011,
      "learning_rate": 0.00018515159038605828,
      "loss": 1.0393,
      "step": 2781
    },
    {
      "epoch": 0.20053341022129315,
      "grad_norm": 2.0699498460079915,
      "learning_rate": 0.00018513934648223937,
      "loss": 1.2016,
      "step": 2782
    },
    {
      "epoch": 0.20060549268363007,
      "grad_norm": 0.5807162740872107,
      "learning_rate": 0.0001851270979375644,
      "loss": 1.0717,
      "step": 2783
    },
    {
      "epoch": 0.20067757514596699,
      "grad_norm": 1.004962414116413,
      "learning_rate": 0.000185114844752701,
      "loss": 1.2118,
      "step": 2784
    },
    {
      "epoch": 0.2007496576083039,
      "grad_norm": 0.29334133463362366,
      "learning_rate": 0.00018510258692831712,
      "loss": 1.074,
      "step": 2785
    },
    {
      "epoch": 0.20082174007064083,
      "grad_norm": 0.17441483342536698,
      "learning_rate": 0.00018509032446508092,
      "loss": 1.0323,
      "step": 2786
    },
    {
      "epoch": 0.20089382253297772,
      "grad_norm": 0.20332838155760524,
      "learning_rate": 0.00018507805736366078,
      "loss": 1.0571,
      "step": 2787
    },
    {
      "epoch": 0.20096590499531464,
      "grad_norm": 2.1144768515348114,
      "learning_rate": 0.00018506578562472537,
      "loss": 1.0508,
      "step": 2788
    },
    {
      "epoch": 0.20103798745765156,
      "grad_norm": 0.6492652440366876,
      "learning_rate": 0.00018505350924894365,
      "loss": 1.1674,
      "step": 2789
    },
    {
      "epoch": 0.20111006991998848,
      "grad_norm": 1.4283440090103923,
      "learning_rate": 0.00018504122823698472,
      "loss": 1.1089,
      "step": 2790
    },
    {
      "epoch": 0.20118215238232537,
      "grad_norm": 0.3261791976492944,
      "learning_rate": 0.0001850289425895181,
      "loss": 1.1116,
      "step": 2791
    },
    {
      "epoch": 0.2012542348446623,
      "grad_norm": 0.15440022622924138,
      "learning_rate": 0.00018501665230721338,
      "loss": 1.1351,
      "step": 2792
    },
    {
      "epoch": 0.2013263173069992,
      "grad_norm": 0.23675012574867668,
      "learning_rate": 0.00018500435739074057,
      "loss": 1.0782,
      "step": 2793
    },
    {
      "epoch": 0.20139839976933613,
      "grad_norm": 0.20789992070027707,
      "learning_rate": 0.00018499205784076978,
      "loss": 1.126,
      "step": 2794
    },
    {
      "epoch": 0.20147048223167302,
      "grad_norm": 0.1758012267736287,
      "learning_rate": 0.0001849797536579715,
      "loss": 1.081,
      "step": 2795
    },
    {
      "epoch": 0.20154256469400994,
      "grad_norm": 0.13154546889023547,
      "learning_rate": 0.0001849674448430164,
      "loss": 1.0428,
      "step": 2796
    },
    {
      "epoch": 0.20161464715634686,
      "grad_norm": 0.24763977322730654,
      "learning_rate": 0.0001849551313965754,
      "loss": 1.1036,
      "step": 2797
    },
    {
      "epoch": 0.20168672961868378,
      "grad_norm": 2.4908053680066793,
      "learning_rate": 0.0001849428133193197,
      "loss": 1.1247,
      "step": 2798
    },
    {
      "epoch": 0.2017588120810207,
      "grad_norm": 0.11370029209956933,
      "learning_rate": 0.00018493049061192072,
      "loss": 1.0609,
      "step": 2799
    },
    {
      "epoch": 0.2018308945433576,
      "grad_norm": 0.19015107158051045,
      "learning_rate": 0.00018491816327505025,
      "loss": 1.1178,
      "step": 2800
    },
    {
      "epoch": 0.2019029770056945,
      "grad_norm": 0.3680841581233173,
      "learning_rate": 0.00018490583130938016,
      "loss": 1.0674,
      "step": 2801
    },
    {
      "epoch": 0.20197505946803143,
      "grad_norm": 0.12362788710078085,
      "learning_rate": 0.00018489349471558267,
      "loss": 1.0354,
      "step": 2802
    },
    {
      "epoch": 0.20204714193036835,
      "grad_norm": 0.16168085379826266,
      "learning_rate": 0.00018488115349433025,
      "loss": 1.0683,
      "step": 2803
    },
    {
      "epoch": 0.20211922439270524,
      "grad_norm": 0.22563004929175984,
      "learning_rate": 0.0001848688076462956,
      "loss": 1.1168,
      "step": 2804
    },
    {
      "epoch": 0.20219130685504216,
      "grad_norm": 0.12133855369093641,
      "learning_rate": 0.00018485645717215168,
      "loss": 1.0886,
      "step": 2805
    },
    {
      "epoch": 0.20226338931737908,
      "grad_norm": 0.29158684000599167,
      "learning_rate": 0.00018484410207257168,
      "loss": 1.0951,
      "step": 2806
    },
    {
      "epoch": 0.202335471779716,
      "grad_norm": 0.21996399073204853,
      "learning_rate": 0.00018483174234822908,
      "loss": 1.0507,
      "step": 2807
    },
    {
      "epoch": 0.2024075542420529,
      "grad_norm": 0.12637813000786766,
      "learning_rate": 0.0001848193779997976,
      "loss": 1.0137,
      "step": 2808
    },
    {
      "epoch": 0.20247963670438981,
      "grad_norm": 0.5337284486971456,
      "learning_rate": 0.0001848070090279512,
      "loss": 1.0695,
      "step": 2809
    },
    {
      "epoch": 0.20255171916672673,
      "grad_norm": 0.1497767564779832,
      "learning_rate": 0.0001847946354333641,
      "loss": 1.1475,
      "step": 2810
    },
    {
      "epoch": 0.20262380162906365,
      "grad_norm": 0.1348213681228322,
      "learning_rate": 0.00018478225721671077,
      "loss": 1.1036,
      "step": 2811
    },
    {
      "epoch": 0.20269588409140057,
      "grad_norm": 0.2796644191194822,
      "learning_rate": 0.00018476987437866598,
      "loss": 1.0033,
      "step": 2812
    },
    {
      "epoch": 0.20276796655373747,
      "grad_norm": 0.22702024521652775,
      "learning_rate": 0.00018475748691990462,
      "loss": 1.0902,
      "step": 2813
    },
    {
      "epoch": 0.2028400490160744,
      "grad_norm": 0.21064361907325052,
      "learning_rate": 0.00018474509484110198,
      "loss": 1.1404,
      "step": 2814
    },
    {
      "epoch": 0.2029121314784113,
      "grad_norm": 0.49698686301420525,
      "learning_rate": 0.0001847326981429335,
      "loss": 1.026,
      "step": 2815
    },
    {
      "epoch": 0.20298421394074823,
      "grad_norm": 0.13801555526441192,
      "learning_rate": 0.00018472029682607496,
      "loss": 1.0425,
      "step": 2816
    },
    {
      "epoch": 0.20305629640308512,
      "grad_norm": 0.18781268249347569,
      "learning_rate": 0.00018470789089120227,
      "loss": 1.0358,
      "step": 2817
    },
    {
      "epoch": 0.20312837886542204,
      "grad_norm": 0.16191329762537685,
      "learning_rate": 0.00018469548033899175,
      "loss": 1.0293,
      "step": 2818
    },
    {
      "epoch": 0.20320046132775896,
      "grad_norm": 0.1501411778718163,
      "learning_rate": 0.00018468306517011984,
      "loss": 1.1039,
      "step": 2819
    },
    {
      "epoch": 0.20327254379009588,
      "grad_norm": 0.16862695293724783,
      "learning_rate": 0.00018467064538526328,
      "loss": 1.1424,
      "step": 2820
    },
    {
      "epoch": 0.20334462625243277,
      "grad_norm": 0.12827265719813846,
      "learning_rate": 0.00018465822098509904,
      "loss": 1.1089,
      "step": 2821
    },
    {
      "epoch": 0.2034167087147697,
      "grad_norm": 0.1472279177795062,
      "learning_rate": 0.0001846457919703044,
      "loss": 1.1344,
      "step": 2822
    },
    {
      "epoch": 0.2034887911771066,
      "grad_norm": 0.24372633584116618,
      "learning_rate": 0.00018463335834155683,
      "loss": 0.9881,
      "step": 2823
    },
    {
      "epoch": 0.20356087363944353,
      "grad_norm": 0.12342956899551975,
      "learning_rate": 0.00018462092009953408,
      "loss": 1.0557,
      "step": 2824
    },
    {
      "epoch": 0.20363295610178045,
      "grad_norm": 0.1288395555513728,
      "learning_rate": 0.00018460847724491414,
      "loss": 1.0027,
      "step": 2825
    },
    {
      "epoch": 0.20370503856411734,
      "grad_norm": 0.14666799786680593,
      "learning_rate": 0.00018459602977837527,
      "loss": 1.0425,
      "step": 2826
    },
    {
      "epoch": 0.20377712102645426,
      "grad_norm": 0.15153472948852648,
      "learning_rate": 0.00018458357770059593,
      "loss": 1.137,
      "step": 2827
    },
    {
      "epoch": 0.20384920348879118,
      "grad_norm": 0.15903867289431775,
      "learning_rate": 0.00018457112101225492,
      "loss": 1.0725,
      "step": 2828
    },
    {
      "epoch": 0.2039212859511281,
      "grad_norm": 0.13896473012531366,
      "learning_rate": 0.00018455865971403124,
      "loss": 1.0901,
      "step": 2829
    },
    {
      "epoch": 0.203993368413465,
      "grad_norm": 0.8316226527254205,
      "learning_rate": 0.0001845461938066041,
      "loss": 1.1155,
      "step": 2830
    },
    {
      "epoch": 0.2040654508758019,
      "grad_norm": 0.1807651086406401,
      "learning_rate": 0.00018453372329065303,
      "loss": 1.058,
      "step": 2831
    },
    {
      "epoch": 0.20413753333813883,
      "grad_norm": 0.2120735341873611,
      "learning_rate": 0.00018452124816685777,
      "loss": 1.0872,
      "step": 2832
    },
    {
      "epoch": 0.20420961580047575,
      "grad_norm": 0.1786743997110807,
      "learning_rate": 0.00018450876843589836,
      "loss": 1.0343,
      "step": 2833
    },
    {
      "epoch": 0.20428169826281264,
      "grad_norm": 0.15780238550843956,
      "learning_rate": 0.000184496284098455,
      "loss": 1.0859,
      "step": 2834
    },
    {
      "epoch": 0.20435378072514956,
      "grad_norm": 0.1471964129342914,
      "learning_rate": 0.00018448379515520827,
      "loss": 1.0205,
      "step": 2835
    },
    {
      "epoch": 0.20442586318748648,
      "grad_norm": 1.3540650735854163,
      "learning_rate": 0.00018447130160683886,
      "loss": 1.1384,
      "step": 2836
    },
    {
      "epoch": 0.2044979456498234,
      "grad_norm": 0.4116307025262551,
      "learning_rate": 0.0001844588034540278,
      "loss": 1.0937,
      "step": 2837
    },
    {
      "epoch": 0.20457002811216032,
      "grad_norm": 0.16521034455582165,
      "learning_rate": 0.0001844463006974564,
      "loss": 1.0348,
      "step": 2838
    },
    {
      "epoch": 0.20464211057449722,
      "grad_norm": 0.1365905122890847,
      "learning_rate": 0.00018443379333780613,
      "loss": 1.0552,
      "step": 2839
    },
    {
      "epoch": 0.20471419303683414,
      "grad_norm": 2.03533061005461,
      "learning_rate": 0.00018442128137575873,
      "loss": 1.0816,
      "step": 2840
    },
    {
      "epoch": 0.20478627549917106,
      "grad_norm": 7.486576678421146,
      "learning_rate": 0.00018440876481199626,
      "loss": 1.148,
      "step": 2841
    },
    {
      "epoch": 0.20485835796150798,
      "grad_norm": 0.38935822278655036,
      "learning_rate": 0.00018439624364720097,
      "loss": 1.0597,
      "step": 2842
    },
    {
      "epoch": 0.20493044042384487,
      "grad_norm": 0.5302937468623196,
      "learning_rate": 0.00018438371788205536,
      "loss": 1.0667,
      "step": 2843
    },
    {
      "epoch": 0.2050025228861818,
      "grad_norm": 0.9948246224812177,
      "learning_rate": 0.0001843711875172422,
      "loss": 1.0811,
      "step": 2844
    },
    {
      "epoch": 0.2050746053485187,
      "grad_norm": 1.1929652821505967,
      "learning_rate": 0.00018435865255344452,
      "loss": 1.0982,
      "step": 2845
    },
    {
      "epoch": 0.20514668781085563,
      "grad_norm": 1.0158624198998427,
      "learning_rate": 0.00018434611299134558,
      "loss": 1.1162,
      "step": 2846
    },
    {
      "epoch": 0.20521877027319252,
      "grad_norm": 0.4524489724421627,
      "learning_rate": 0.00018433356883162892,
      "loss": 1.1268,
      "step": 2847
    },
    {
      "epoch": 0.20529085273552944,
      "grad_norm": 0.7249192762373613,
      "learning_rate": 0.00018432102007497826,
      "loss": 1.0173,
      "step": 2848
    },
    {
      "epoch": 0.20536293519786636,
      "grad_norm": 1.6897547156108592,
      "learning_rate": 0.00018430846672207768,
      "loss": 1.1221,
      "step": 2849
    },
    {
      "epoch": 0.20543501766020328,
      "grad_norm": 0.29823068506172834,
      "learning_rate": 0.0001842959087736114,
      "loss": 1.1057,
      "step": 2850
    },
    {
      "epoch": 0.2055071001225402,
      "grad_norm": 0.18437945746883191,
      "learning_rate": 0.00018428334623026395,
      "loss": 1.106,
      "step": 2851
    },
    {
      "epoch": 0.2055791825848771,
      "grad_norm": 0.13653430977096503,
      "learning_rate": 0.0001842707790927201,
      "loss": 1.0589,
      "step": 2852
    },
    {
      "epoch": 0.205651265047214,
      "grad_norm": 1.8460592182325992,
      "learning_rate": 0.0001842582073616649,
      "loss": 1.0277,
      "step": 2853
    },
    {
      "epoch": 0.20572334750955093,
      "grad_norm": 3.1728644136508763,
      "learning_rate": 0.00018424563103778364,
      "loss": 1.1823,
      "step": 2854
    },
    {
      "epoch": 0.20579542997188785,
      "grad_norm": 0.5990390677977431,
      "learning_rate": 0.00018423305012176176,
      "loss": 1.0418,
      "step": 2855
    },
    {
      "epoch": 0.20586751243422474,
      "grad_norm": 2.0384240315059907,
      "learning_rate": 0.00018422046461428508,
      "loss": 1.0999,
      "step": 2856
    },
    {
      "epoch": 0.20593959489656166,
      "grad_norm": 7.249656338838709,
      "learning_rate": 0.00018420787451603961,
      "loss": 1.1408,
      "step": 2857
    },
    {
      "epoch": 0.20601167735889858,
      "grad_norm": 0.2845039578094495,
      "learning_rate": 0.00018419527982771166,
      "loss": 1.0942,
      "step": 2858
    },
    {
      "epoch": 0.2060837598212355,
      "grad_norm": 0.5534498433250252,
      "learning_rate": 0.00018418268054998772,
      "loss": 1.1281,
      "step": 2859
    },
    {
      "epoch": 0.2061558422835724,
      "grad_norm": 1.6355607226521525,
      "learning_rate": 0.00018417007668355457,
      "loss": 1.1003,
      "step": 2860
    },
    {
      "epoch": 0.2062279247459093,
      "grad_norm": 0.6204430997582823,
      "learning_rate": 0.00018415746822909922,
      "loss": 1.1074,
      "step": 2861
    },
    {
      "epoch": 0.20630000720824623,
      "grad_norm": 1.2132957277401888,
      "learning_rate": 0.000184144855187309,
      "loss": 1.1826,
      "step": 2862
    },
    {
      "epoch": 0.20637208967058315,
      "grad_norm": 5.052660865667796,
      "learning_rate": 0.00018413223755887136,
      "loss": 1.2622,
      "step": 2863
    },
    {
      "epoch": 0.20644417213292007,
      "grad_norm": 0.8858023454450196,
      "learning_rate": 0.0001841196153444741,
      "loss": 1.212,
      "step": 2864
    },
    {
      "epoch": 0.20651625459525697,
      "grad_norm": 0.8484839799107246,
      "learning_rate": 0.00018410698854480525,
      "loss": 1.1801,
      "step": 2865
    },
    {
      "epoch": 0.20658833705759388,
      "grad_norm": 2.4093093918048556,
      "learning_rate": 0.00018409435716055308,
      "loss": 1.0649,
      "step": 2866
    },
    {
      "epoch": 0.2066604195199308,
      "grad_norm": 0.7938308626722805,
      "learning_rate": 0.00018408172119240614,
      "loss": 1.1956,
      "step": 2867
    },
    {
      "epoch": 0.20673250198226772,
      "grad_norm": 0.44586495107939156,
      "learning_rate": 0.00018406908064105314,
      "loss": 1.2778,
      "step": 2868
    },
    {
      "epoch": 0.20680458444460462,
      "grad_norm": 0.4684623852564254,
      "learning_rate": 0.00018405643550718317,
      "loss": 1.0851,
      "step": 2869
    },
    {
      "epoch": 0.20687666690694154,
      "grad_norm": 0.3001495590172088,
      "learning_rate": 0.00018404378579148546,
      "loss": 1.0785,
      "step": 2870
    },
    {
      "epoch": 0.20694874936927846,
      "grad_norm": 0.8472366284367242,
      "learning_rate": 0.00018403113149464953,
      "loss": 1.2154,
      "step": 2871
    },
    {
      "epoch": 0.20702083183161538,
      "grad_norm": 0.3296390524542271,
      "learning_rate": 0.00018401847261736523,
      "loss": 1.1574,
      "step": 2872
    },
    {
      "epoch": 0.20709291429395227,
      "grad_norm": 0.9221157896303774,
      "learning_rate": 0.00018400580916032246,
      "loss": 1.1991,
      "step": 2873
    },
    {
      "epoch": 0.2071649967562892,
      "grad_norm": 0.6323573177191439,
      "learning_rate": 0.0001839931411242116,
      "loss": 1.1868,
      "step": 2874
    },
    {
      "epoch": 0.2072370792186261,
      "grad_norm": 0.29132646966792614,
      "learning_rate": 0.0001839804685097231,
      "loss": 1.1736,
      "step": 2875
    },
    {
      "epoch": 0.20730916168096303,
      "grad_norm": 0.37331128919820905,
      "learning_rate": 0.00018396779131754775,
      "loss": 1.1402,
      "step": 2876
    },
    {
      "epoch": 0.20738124414329995,
      "grad_norm": 0.2565463283558106,
      "learning_rate": 0.00018395510954837658,
      "loss": 1.0114,
      "step": 2877
    },
    {
      "epoch": 0.20745332660563684,
      "grad_norm": 0.8223359747249567,
      "learning_rate": 0.00018394242320290087,
      "loss": 1.1696,
      "step": 2878
    },
    {
      "epoch": 0.20752540906797376,
      "grad_norm": 0.240674607369847,
      "learning_rate": 0.00018392973228181209,
      "loss": 1.0943,
      "step": 2879
    },
    {
      "epoch": 0.20759749153031068,
      "grad_norm": 1.9691483342384009,
      "learning_rate": 0.00018391703678580207,
      "loss": 1.0985,
      "step": 2880
    },
    {
      "epoch": 0.2076695739926476,
      "grad_norm": 0.24675049881675665,
      "learning_rate": 0.0001839043367155628,
      "loss": 1.0849,
      "step": 2881
    },
    {
      "epoch": 0.2077416564549845,
      "grad_norm": 0.5081664865969457,
      "learning_rate": 0.00018389163207178656,
      "loss": 1.1477,
      "step": 2882
    },
    {
      "epoch": 0.2078137389173214,
      "grad_norm": 0.38156511762489653,
      "learning_rate": 0.00018387892285516582,
      "loss": 1.0981,
      "step": 2883
    },
    {
      "epoch": 0.20788582137965833,
      "grad_norm": 2.0482489430194,
      "learning_rate": 0.00018386620906639343,
      "loss": 1.0605,
      "step": 2884
    },
    {
      "epoch": 0.20795790384199525,
      "grad_norm": 0.25174753749451056,
      "learning_rate": 0.00018385349070616233,
      "loss": 1.1658,
      "step": 2885
    },
    {
      "epoch": 0.20802998630433214,
      "grad_norm": 1.0836038143349844,
      "learning_rate": 0.00018384076777516582,
      "loss": 1.1464,
      "step": 2886
    },
    {
      "epoch": 0.20810206876666906,
      "grad_norm": 0.247403896876518,
      "learning_rate": 0.0001838280402740974,
      "loss": 1.1198,
      "step": 2887
    },
    {
      "epoch": 0.20817415122900598,
      "grad_norm": 0.7771850885587468,
      "learning_rate": 0.00018381530820365085,
      "loss": 1.182,
      "step": 2888
    },
    {
      "epoch": 0.2082462336913429,
      "grad_norm": 0.3695551525468368,
      "learning_rate": 0.00018380257156452013,
      "loss": 1.1633,
      "step": 2889
    },
    {
      "epoch": 0.20831831615367982,
      "grad_norm": 0.22839575946762647,
      "learning_rate": 0.00018378983035739958,
      "loss": 1.0898,
      "step": 2890
    },
    {
      "epoch": 0.20839039861601671,
      "grad_norm": 0.3993392193858924,
      "learning_rate": 0.0001837770845829837,
      "loss": 1.1696,
      "step": 2891
    },
    {
      "epoch": 0.20846248107835363,
      "grad_norm": 0.1520229729272399,
      "learning_rate": 0.00018376433424196716,
      "loss": 1.0584,
      "step": 2892
    },
    {
      "epoch": 0.20853456354069055,
      "grad_norm": 0.5319298910349398,
      "learning_rate": 0.00018375157933504503,
      "loss": 1.0621,
      "step": 2893
    },
    {
      "epoch": 0.20860664600302747,
      "grad_norm": 0.46040790867533704,
      "learning_rate": 0.00018373881986291258,
      "loss": 1.0522,
      "step": 2894
    },
    {
      "epoch": 0.20867872846536437,
      "grad_norm": 0.5393987123892506,
      "learning_rate": 0.0001837260558262653,
      "loss": 1.1213,
      "step": 2895
    },
    {
      "epoch": 0.20875081092770129,
      "grad_norm": 0.23793483163574816,
      "learning_rate": 0.00018371328722579895,
      "loss": 0.9579,
      "step": 2896
    },
    {
      "epoch": 0.2088228933900382,
      "grad_norm": 0.5132862281283631,
      "learning_rate": 0.0001837005140622095,
      "loss": 1.1051,
      "step": 2897
    },
    {
      "epoch": 0.20889497585237513,
      "grad_norm": 0.5108390443494113,
      "learning_rate": 0.00018368773633619325,
      "loss": 1.1277,
      "step": 2898
    },
    {
      "epoch": 0.20896705831471202,
      "grad_norm": 0.3266780720337394,
      "learning_rate": 0.00018367495404844668,
      "loss": 1.1323,
      "step": 2899
    },
    {
      "epoch": 0.20903914077704894,
      "grad_norm": 0.27854595626336465,
      "learning_rate": 0.0001836621671996665,
      "loss": 1.178,
      "step": 2900
    },
    {
      "epoch": 0.20911122323938586,
      "grad_norm": 0.18247082152426536,
      "learning_rate": 0.0001836493757905498,
      "loss": 1.1729,
      "step": 2901
    },
    {
      "epoch": 0.20918330570172278,
      "grad_norm": 0.14474996383916192,
      "learning_rate": 0.00018363657982179373,
      "loss": 1.1056,
      "step": 2902
    },
    {
      "epoch": 0.2092553881640597,
      "grad_norm": 0.8736685375007156,
      "learning_rate": 0.00018362377929409583,
      "loss": 1.0895,
      "step": 2903
    },
    {
      "epoch": 0.2093274706263966,
      "grad_norm": 0.31343656942724746,
      "learning_rate": 0.00018361097420815386,
      "loss": 1.1124,
      "step": 2904
    },
    {
      "epoch": 0.2093995530887335,
      "grad_norm": 0.20168006393440677,
      "learning_rate": 0.00018359816456466575,
      "loss": 1.1137,
      "step": 2905
    },
    {
      "epoch": 0.20947163555107043,
      "grad_norm": 0.379674038417508,
      "learning_rate": 0.00018358535036432983,
      "loss": 1.1441,
      "step": 2906
    },
    {
      "epoch": 0.20954371801340735,
      "grad_norm": 0.2776471490293574,
      "learning_rate": 0.0001835725316078445,
      "loss": 1.0985,
      "step": 2907
    },
    {
      "epoch": 0.20961580047574424,
      "grad_norm": 0.40737434898077823,
      "learning_rate": 0.00018355970829590857,
      "loss": 1.1309,
      "step": 2908
    },
    {
      "epoch": 0.20968788293808116,
      "grad_norm": 0.18634070105871992,
      "learning_rate": 0.00018354688042922096,
      "loss": 1.0353,
      "step": 2909
    },
    {
      "epoch": 0.20975996540041808,
      "grad_norm": 0.23425676917635957,
      "learning_rate": 0.00018353404800848096,
      "loss": 1.115,
      "step": 2910
    },
    {
      "epoch": 0.209832047862755,
      "grad_norm": 0.4970940400690022,
      "learning_rate": 0.000183521211034388,
      "loss": 1.1229,
      "step": 2911
    },
    {
      "epoch": 0.2099041303250919,
      "grad_norm": 0.2531833936158712,
      "learning_rate": 0.00018350836950764188,
      "loss": 1.1576,
      "step": 2912
    },
    {
      "epoch": 0.2099762127874288,
      "grad_norm": 0.18416864355999035,
      "learning_rate": 0.0001834955234289425,
      "loss": 1.1276,
      "step": 2913
    },
    {
      "epoch": 0.21004829524976573,
      "grad_norm": 0.2763313168031476,
      "learning_rate": 0.00018348267279899013,
      "loss": 1.0824,
      "step": 2914
    },
    {
      "epoch": 0.21012037771210265,
      "grad_norm": 0.5583391693313497,
      "learning_rate": 0.00018346981761848522,
      "loss": 1.2781,
      "step": 2915
    },
    {
      "epoch": 0.21019246017443957,
      "grad_norm": 0.1556334219965276,
      "learning_rate": 0.00018345695788812854,
      "loss": 1.079,
      "step": 2916
    },
    {
      "epoch": 0.21026454263677646,
      "grad_norm": 0.25596521649601994,
      "learning_rate": 0.000183444093608621,
      "loss": 1.1665,
      "step": 2917
    },
    {
      "epoch": 0.21033662509911338,
      "grad_norm": 0.28204656433677566,
      "learning_rate": 0.00018343122478066385,
      "loss": 1.0339,
      "step": 2918
    },
    {
      "epoch": 0.2104087075614503,
      "grad_norm": 0.5319352288304344,
      "learning_rate": 0.00018341835140495854,
      "loss": 1.0255,
      "step": 2919
    },
    {
      "epoch": 0.21048079002378722,
      "grad_norm": 1.5992872401059715,
      "learning_rate": 0.00018340547348220682,
      "loss": 1.1206,
      "step": 2920
    },
    {
      "epoch": 0.21055287248612412,
      "grad_norm": 0.8447849585366511,
      "learning_rate": 0.00018339259101311063,
      "loss": 1.0785,
      "step": 2921
    },
    {
      "epoch": 0.21062495494846104,
      "grad_norm": 0.2933614042378728,
      "learning_rate": 0.0001833797039983722,
      "loss": 1.0522,
      "step": 2922
    },
    {
      "epoch": 0.21069703741079795,
      "grad_norm": 0.29802799078052045,
      "learning_rate": 0.00018336681243869392,
      "loss": 1.0618,
      "step": 2923
    },
    {
      "epoch": 0.21076911987313487,
      "grad_norm": 0.2785176358977195,
      "learning_rate": 0.00018335391633477857,
      "loss": 1.0917,
      "step": 2924
    },
    {
      "epoch": 0.21084120233547177,
      "grad_norm": 0.2813453481790206,
      "learning_rate": 0.00018334101568732905,
      "loss": 1.1633,
      "step": 2925
    },
    {
      "epoch": 0.2109132847978087,
      "grad_norm": 0.6534565052031258,
      "learning_rate": 0.00018332811049704862,
      "loss": 1.0692,
      "step": 2926
    },
    {
      "epoch": 0.2109853672601456,
      "grad_norm": 0.21845595004667487,
      "learning_rate": 0.00018331520076464065,
      "loss": 1.0981,
      "step": 2927
    },
    {
      "epoch": 0.21105744972248253,
      "grad_norm": 0.1820302278543949,
      "learning_rate": 0.0001833022864908089,
      "loss": 1.1259,
      "step": 2928
    },
    {
      "epoch": 0.21112953218481945,
      "grad_norm": 0.24186973943342172,
      "learning_rate": 0.00018328936767625728,
      "loss": 1.0865,
      "step": 2929
    },
    {
      "epoch": 0.21120161464715634,
      "grad_norm": 0.21974901858084706,
      "learning_rate": 0.00018327644432169002,
      "loss": 1.0777,
      "step": 2930
    },
    {
      "epoch": 0.21127369710949326,
      "grad_norm": 1.0144946477453747,
      "learning_rate": 0.0001832635164278115,
      "loss": 1.1322,
      "step": 2931
    },
    {
      "epoch": 0.21134577957183018,
      "grad_norm": 0.31360003946967513,
      "learning_rate": 0.00018325058399532647,
      "loss": 1.1095,
      "step": 2932
    },
    {
      "epoch": 0.2114178620341671,
      "grad_norm": 0.4988565724814746,
      "learning_rate": 0.0001832376470249398,
      "loss": 1.0935,
      "step": 2933
    },
    {
      "epoch": 0.211489944496504,
      "grad_norm": 0.3893174511662761,
      "learning_rate": 0.00018322470551735673,
      "loss": 0.9783,
      "step": 2934
    },
    {
      "epoch": 0.2115620269588409,
      "grad_norm": 0.23722951705408882,
      "learning_rate": 0.00018321175947328268,
      "loss": 1.0814,
      "step": 2935
    },
    {
      "epoch": 0.21163410942117783,
      "grad_norm": 0.3564922034448822,
      "learning_rate": 0.00018319880889342325,
      "loss": 1.1117,
      "step": 2936
    },
    {
      "epoch": 0.21170619188351475,
      "grad_norm": 0.31819005019562474,
      "learning_rate": 0.00018318585377848442,
      "loss": 1.0986,
      "step": 2937
    },
    {
      "epoch": 0.21177827434585164,
      "grad_norm": 0.22744651265224908,
      "learning_rate": 0.0001831728941291724,
      "loss": 1.1169,
      "step": 2938
    },
    {
      "epoch": 0.21185035680818856,
      "grad_norm": 0.19499893593851275,
      "learning_rate": 0.00018315992994619355,
      "loss": 1.1498,
      "step": 2939
    },
    {
      "epoch": 0.21192243927052548,
      "grad_norm": 0.1385822412746984,
      "learning_rate": 0.00018314696123025454,
      "loss": 1.0166,
      "step": 2940
    },
    {
      "epoch": 0.2119945217328624,
      "grad_norm": 0.17814769563033075,
      "learning_rate": 0.0001831339879820623,
      "loss": 1.0889,
      "step": 2941
    },
    {
      "epoch": 0.21206660419519932,
      "grad_norm": 0.2677361712578228,
      "learning_rate": 0.00018312101020232394,
      "loss": 1.0497,
      "step": 2942
    },
    {
      "epoch": 0.2121386866575362,
      "grad_norm": 0.15313685193502385,
      "learning_rate": 0.00018310802789174693,
      "loss": 1.1336,
      "step": 2943
    },
    {
      "epoch": 0.21221076911987313,
      "grad_norm": 0.28788225736556045,
      "learning_rate": 0.0001830950410510389,
      "loss": 1.0416,
      "step": 2944
    },
    {
      "epoch": 0.21228285158221005,
      "grad_norm": 0.16137311167324878,
      "learning_rate": 0.00018308204968090774,
      "loss": 1.0493,
      "step": 2945
    },
    {
      "epoch": 0.21235493404454697,
      "grad_norm": 0.25154703782600024,
      "learning_rate": 0.00018306905378206157,
      "loss": 1.1928,
      "step": 2946
    },
    {
      "epoch": 0.21242701650688386,
      "grad_norm": 0.3653985244476339,
      "learning_rate": 0.00018305605335520886,
      "loss": 1.0888,
      "step": 2947
    },
    {
      "epoch": 0.21249909896922078,
      "grad_norm": 0.28875230681193653,
      "learning_rate": 0.00018304304840105817,
      "loss": 1.0924,
      "step": 2948
    },
    {
      "epoch": 0.2125711814315577,
      "grad_norm": 0.2831525327608258,
      "learning_rate": 0.00018303003892031843,
      "loss": 1.0988,
      "step": 2949
    },
    {
      "epoch": 0.21264326389389462,
      "grad_norm": 0.15658439238172353,
      "learning_rate": 0.00018301702491369876,
      "loss": 1.0368,
      "step": 2950
    },
    {
      "epoch": 0.21271534635623152,
      "grad_norm": 0.18451122034374293,
      "learning_rate": 0.00018300400638190855,
      "loss": 1.1081,
      "step": 2951
    },
    {
      "epoch": 0.21278742881856844,
      "grad_norm": 1.6845476745312482,
      "learning_rate": 0.0001829909833256574,
      "loss": 1.0582,
      "step": 2952
    },
    {
      "epoch": 0.21285951128090536,
      "grad_norm": 0.21581472969042578,
      "learning_rate": 0.0001829779557456552,
      "loss": 1.0682,
      "step": 2953
    },
    {
      "epoch": 0.21293159374324228,
      "grad_norm": 0.3378251853042997,
      "learning_rate": 0.00018296492364261208,
      "loss": 1.1149,
      "step": 2954
    },
    {
      "epoch": 0.2130036762055792,
      "grad_norm": 0.5392344861059859,
      "learning_rate": 0.00018295188701723838,
      "loss": 1.0709,
      "step": 2955
    },
    {
      "epoch": 0.2130757586679161,
      "grad_norm": 0.553215663613889,
      "learning_rate": 0.00018293884587024475,
      "loss": 1.0781,
      "step": 2956
    },
    {
      "epoch": 0.213147841130253,
      "grad_norm": 0.7619201994450914,
      "learning_rate": 0.000182925800202342,
      "loss": 1.054,
      "step": 2957
    },
    {
      "epoch": 0.21321992359258993,
      "grad_norm": 0.1360861204633599,
      "learning_rate": 0.0001829127500142413,
      "loss": 1.1301,
      "step": 2958
    },
    {
      "epoch": 0.21329200605492685,
      "grad_norm": 1.0426005452962503,
      "learning_rate": 0.00018289969530665395,
      "loss": 1.1332,
      "step": 2959
    },
    {
      "epoch": 0.21336408851726374,
      "grad_norm": 0.15027762924176874,
      "learning_rate": 0.00018288663608029157,
      "loss": 1.0658,
      "step": 2960
    },
    {
      "epoch": 0.21343617097960066,
      "grad_norm": 0.5689820692037321,
      "learning_rate": 0.00018287357233586597,
      "loss": 1.1596,
      "step": 2961
    },
    {
      "epoch": 0.21350825344193758,
      "grad_norm": 0.36607478430090057,
      "learning_rate": 0.0001828605040740893,
      "loss": 1.1364,
      "step": 2962
    },
    {
      "epoch": 0.2135803359042745,
      "grad_norm": 0.3569255550408155,
      "learning_rate": 0.00018284743129567388,
      "loss": 1.0663,
      "step": 2963
    },
    {
      "epoch": 0.2136524183666114,
      "grad_norm": 0.15365830548442574,
      "learning_rate": 0.00018283435400133225,
      "loss": 1.064,
      "step": 2964
    },
    {
      "epoch": 0.2137245008289483,
      "grad_norm": 0.3607815735897181,
      "learning_rate": 0.00018282127219177726,
      "loss": 1.0366,
      "step": 2965
    },
    {
      "epoch": 0.21379658329128523,
      "grad_norm": 0.1543137714257858,
      "learning_rate": 0.00018280818586772203,
      "loss": 0.9749,
      "step": 2966
    },
    {
      "epoch": 0.21386866575362215,
      "grad_norm": 0.2195075085221088,
      "learning_rate": 0.00018279509502987983,
      "loss": 1.1165,
      "step": 2967
    },
    {
      "epoch": 0.21394074821595907,
      "grad_norm": 0.2969208386734238,
      "learning_rate": 0.00018278199967896423,
      "loss": 1.0643,
      "step": 2968
    },
    {
      "epoch": 0.21401283067829596,
      "grad_norm": 0.2168056631114292,
      "learning_rate": 0.00018276889981568906,
      "loss": 1.1837,
      "step": 2969
    },
    {
      "epoch": 0.21408491314063288,
      "grad_norm": 0.37562783529884697,
      "learning_rate": 0.0001827557954407684,
      "loss": 1.0643,
      "step": 2970
    },
    {
      "epoch": 0.2141569956029698,
      "grad_norm": 0.20785889852959977,
      "learning_rate": 0.00018274268655491652,
      "loss": 1.1875,
      "step": 2971
    },
    {
      "epoch": 0.21422907806530672,
      "grad_norm": 0.2790400987138996,
      "learning_rate": 0.000182729573158848,
      "loss": 1.0787,
      "step": 2972
    },
    {
      "epoch": 0.2143011605276436,
      "grad_norm": 0.1770765385155145,
      "learning_rate": 0.0001827164552532776,
      "loss": 1.0439,
      "step": 2973
    },
    {
      "epoch": 0.21437324298998053,
      "grad_norm": 0.1734468837998761,
      "learning_rate": 0.0001827033328389204,
      "loss": 1.1187,
      "step": 2974
    },
    {
      "epoch": 0.21444532545231745,
      "grad_norm": 0.41058743214759336,
      "learning_rate": 0.00018269020591649168,
      "loss": 1.1391,
      "step": 2975
    },
    {
      "epoch": 0.21451740791465437,
      "grad_norm": 0.2014234844097576,
      "learning_rate": 0.00018267707448670696,
      "loss": 1.0985,
      "step": 2976
    },
    {
      "epoch": 0.21458949037699127,
      "grad_norm": 0.32514026959570813,
      "learning_rate": 0.00018266393855028204,
      "loss": 1.1073,
      "step": 2977
    },
    {
      "epoch": 0.21466157283932819,
      "grad_norm": 0.20098510830638658,
      "learning_rate": 0.00018265079810793293,
      "loss": 1.0511,
      "step": 2978
    },
    {
      "epoch": 0.2147336553016651,
      "grad_norm": 0.6445901156508841,
      "learning_rate": 0.00018263765316037592,
      "loss": 1.12,
      "step": 2979
    },
    {
      "epoch": 0.21480573776400202,
      "grad_norm": 0.2165400823823578,
      "learning_rate": 0.00018262450370832748,
      "loss": 1.0492,
      "step": 2980
    },
    {
      "epoch": 0.21487782022633894,
      "grad_norm": 0.3548914553206725,
      "learning_rate": 0.00018261134975250441,
      "loss": 1.0839,
      "step": 2981
    },
    {
      "epoch": 0.21494990268867584,
      "grad_norm": 0.14035254747439393,
      "learning_rate": 0.00018259819129362375,
      "loss": 1.0939,
      "step": 2982
    },
    {
      "epoch": 0.21502198515101276,
      "grad_norm": 0.37337889577940775,
      "learning_rate": 0.00018258502833240272,
      "loss": 1.1026,
      "step": 2983
    },
    {
      "epoch": 0.21509406761334968,
      "grad_norm": 0.16776951787261102,
      "learning_rate": 0.0001825718608695588,
      "loss": 1.0562,
      "step": 2984
    },
    {
      "epoch": 0.2151661500756866,
      "grad_norm": 0.34222253407508146,
      "learning_rate": 0.00018255868890580977,
      "loss": 1.0715,
      "step": 2985
    },
    {
      "epoch": 0.2152382325380235,
      "grad_norm": 0.28963722020865035,
      "learning_rate": 0.00018254551244187357,
      "loss": 1.1172,
      "step": 2986
    },
    {
      "epoch": 0.2153103150003604,
      "grad_norm": 0.49806312808985503,
      "learning_rate": 0.00018253233147846851,
      "loss": 1.1619,
      "step": 2987
    },
    {
      "epoch": 0.21538239746269733,
      "grad_norm": 0.20804991794983835,
      "learning_rate": 0.00018251914601631302,
      "loss": 1.09,
      "step": 2988
    },
    {
      "epoch": 0.21545447992503425,
      "grad_norm": 0.23000624986271823,
      "learning_rate": 0.0001825059560561258,
      "loss": 1.0462,
      "step": 2989
    },
    {
      "epoch": 0.21552656238737114,
      "grad_norm": 0.12380341571812646,
      "learning_rate": 0.0001824927615986259,
      "loss": 1.086,
      "step": 2990
    },
    {
      "epoch": 0.21559864484970806,
      "grad_norm": 0.5255248707722876,
      "learning_rate": 0.00018247956264453249,
      "loss": 1.1928,
      "step": 2991
    },
    {
      "epoch": 0.21567072731204498,
      "grad_norm": 0.14157053895782132,
      "learning_rate": 0.000182466359194565,
      "loss": 1.0929,
      "step": 2992
    },
    {
      "epoch": 0.2157428097743819,
      "grad_norm": 0.16225234901768185,
      "learning_rate": 0.00018245315124944318,
      "loss": 1.0877,
      "step": 2993
    },
    {
      "epoch": 0.21581489223671882,
      "grad_norm": 0.23538289943291987,
      "learning_rate": 0.00018243993880988696,
      "loss": 1.053,
      "step": 2994
    },
    {
      "epoch": 0.2158869746990557,
      "grad_norm": 0.31628771750530277,
      "learning_rate": 0.00018242672187661658,
      "loss": 1.0792,
      "step": 2995
    },
    {
      "epoch": 0.21595905716139263,
      "grad_norm": 0.36191841577842626,
      "learning_rate": 0.00018241350045035244,
      "loss": 1.0837,
      "step": 2996
    },
    {
      "epoch": 0.21603113962372955,
      "grad_norm": 0.142496796078578,
      "learning_rate": 0.0001824002745318152,
      "loss": 1.0505,
      "step": 2997
    },
    {
      "epoch": 0.21610322208606647,
      "grad_norm": 0.13803830898779468,
      "learning_rate": 0.00018238704412172586,
      "loss": 1.0738,
      "step": 2998
    },
    {
      "epoch": 0.21617530454840336,
      "grad_norm": 0.22942340770410052,
      "learning_rate": 0.00018237380922080554,
      "loss": 1.0065,
      "step": 2999
    },
    {
      "epoch": 0.21624738701074028,
      "grad_norm": 0.3738606934349228,
      "learning_rate": 0.0001823605698297757,
      "loss": 1.1955,
      "step": 3000
    },
    {
      "epoch": 0.2163194694730772,
      "grad_norm": 0.12499737871865814,
      "learning_rate": 0.00018234732594935794,
      "loss": 1.1012,
      "step": 3001
    },
    {
      "epoch": 0.21639155193541412,
      "grad_norm": 0.14311744691373607,
      "learning_rate": 0.00018233407758027424,
      "loss": 0.9905,
      "step": 3002
    },
    {
      "epoch": 0.21646363439775101,
      "grad_norm": 0.4308242193758272,
      "learning_rate": 0.00018232082472324676,
      "loss": 1.1626,
      "step": 3003
    },
    {
      "epoch": 0.21653571686008793,
      "grad_norm": 0.2015706039666184,
      "learning_rate": 0.00018230756737899783,
      "loss": 1.0114,
      "step": 3004
    },
    {
      "epoch": 0.21660779932242485,
      "grad_norm": 0.3251783568785245,
      "learning_rate": 0.00018229430554825017,
      "loss": 1.0137,
      "step": 3005
    },
    {
      "epoch": 0.21667988178476177,
      "grad_norm": 0.12743175755534183,
      "learning_rate": 0.0001822810392317266,
      "loss": 1.0249,
      "step": 3006
    },
    {
      "epoch": 0.2167519642470987,
      "grad_norm": 0.14787749616781223,
      "learning_rate": 0.0001822677684301503,
      "loss": 1.0976,
      "step": 3007
    },
    {
      "epoch": 0.21682404670943559,
      "grad_norm": 0.12825830051293272,
      "learning_rate": 0.00018225449314424462,
      "loss": 1.0131,
      "step": 3008
    },
    {
      "epoch": 0.2168961291717725,
      "grad_norm": 0.15132366580953605,
      "learning_rate": 0.0001822412133747332,
      "loss": 1.095,
      "step": 3009
    },
    {
      "epoch": 0.21696821163410943,
      "grad_norm": 0.12255960804914943,
      "learning_rate": 0.0001822279291223399,
      "loss": 1.0083,
      "step": 3010
    },
    {
      "epoch": 0.21704029409644635,
      "grad_norm": 0.22637667937133613,
      "learning_rate": 0.00018221464038778885,
      "loss": 1.0981,
      "step": 3011
    },
    {
      "epoch": 0.21711237655878324,
      "grad_norm": 0.702251644146088,
      "learning_rate": 0.00018220134717180435,
      "loss": 1.1791,
      "step": 3012
    },
    {
      "epoch": 0.21718445902112016,
      "grad_norm": 0.1448262417365993,
      "learning_rate": 0.00018218804947511106,
      "loss": 1.162,
      "step": 3013
    },
    {
      "epoch": 0.21725654148345708,
      "grad_norm": 0.2053504723944822,
      "learning_rate": 0.00018217474729843378,
      "loss": 1.0697,
      "step": 3014
    },
    {
      "epoch": 0.217328623945794,
      "grad_norm": 1.524690224680317,
      "learning_rate": 0.00018216144064249763,
      "loss": 1.1862,
      "step": 3015
    },
    {
      "epoch": 0.2174007064081309,
      "grad_norm": 0.41998155389817865,
      "learning_rate": 0.00018214812950802792,
      "loss": 1.1361,
      "step": 3016
    },
    {
      "epoch": 0.2174727888704678,
      "grad_norm": 1.530922005757083,
      "learning_rate": 0.00018213481389575025,
      "loss": 1.1007,
      "step": 3017
    },
    {
      "epoch": 0.21754487133280473,
      "grad_norm": 0.19645717073689756,
      "learning_rate": 0.0001821214938063904,
      "loss": 1.047,
      "step": 3018
    },
    {
      "epoch": 0.21761695379514165,
      "grad_norm": 0.2236525755383149,
      "learning_rate": 0.00018210816924067448,
      "loss": 0.9884,
      "step": 3019
    },
    {
      "epoch": 0.21768903625747857,
      "grad_norm": 0.15943856798946482,
      "learning_rate": 0.00018209484019932875,
      "loss": 1.1137,
      "step": 3020
    },
    {
      "epoch": 0.21776111871981546,
      "grad_norm": 0.21987961447150467,
      "learning_rate": 0.0001820815066830798,
      "loss": 1.1225,
      "step": 3021
    },
    {
      "epoch": 0.21783320118215238,
      "grad_norm": 0.28437456092967045,
      "learning_rate": 0.00018206816869265443,
      "loss": 1.1445,
      "step": 3022
    },
    {
      "epoch": 0.2179052836444893,
      "grad_norm": 0.18889883962830312,
      "learning_rate": 0.00018205482622877966,
      "loss": 1.1057,
      "step": 3023
    },
    {
      "epoch": 0.21797736610682622,
      "grad_norm": 0.3525052946885428,
      "learning_rate": 0.0001820414792921828,
      "loss": 1.1003,
      "step": 3024
    },
    {
      "epoch": 0.2180494485691631,
      "grad_norm": 0.18904137450975902,
      "learning_rate": 0.0001820281278835913,
      "loss": 1.0829,
      "step": 3025
    },
    {
      "epoch": 0.21812153103150003,
      "grad_norm": 0.22812670753083325,
      "learning_rate": 0.000182014772003733,
      "loss": 1.0333,
      "step": 3026
    },
    {
      "epoch": 0.21819361349383695,
      "grad_norm": 0.21123610501590556,
      "learning_rate": 0.0001820014116533359,
      "loss": 1.0863,
      "step": 3027
    },
    {
      "epoch": 0.21826569595617387,
      "grad_norm": 0.1565109648079618,
      "learning_rate": 0.00018198804683312826,
      "loss": 1.083,
      "step": 3028
    },
    {
      "epoch": 0.21833777841851076,
      "grad_norm": 0.14024309609683608,
      "learning_rate": 0.00018197467754383858,
      "loss": 1.1389,
      "step": 3029
    },
    {
      "epoch": 0.21840986088084768,
      "grad_norm": 0.15627206817812722,
      "learning_rate": 0.00018196130378619562,
      "loss": 1.0177,
      "step": 3030
    },
    {
      "epoch": 0.2184819433431846,
      "grad_norm": 1.9316112494407423,
      "learning_rate": 0.0001819479255609284,
      "loss": 1.0685,
      "step": 3031
    },
    {
      "epoch": 0.21855402580552152,
      "grad_norm": 0.21850080431781826,
      "learning_rate": 0.00018193454286876602,
      "loss": 1.106,
      "step": 3032
    },
    {
      "epoch": 0.21862610826785844,
      "grad_norm": 0.15199663313372347,
      "learning_rate": 0.0001819211557104381,
      "loss": 1.1063,
      "step": 3033
    },
    {
      "epoch": 0.21869819073019534,
      "grad_norm": 0.16984815071478365,
      "learning_rate": 0.00018190776408667428,
      "loss": 1.0049,
      "step": 3034
    },
    {
      "epoch": 0.21877027319253226,
      "grad_norm": 0.2306368314891108,
      "learning_rate": 0.00018189436799820457,
      "loss": 1.0361,
      "step": 3035
    },
    {
      "epoch": 0.21884235565486918,
      "grad_norm": 0.2427693220198002,
      "learning_rate": 0.00018188096744575918,
      "loss": 1.1594,
      "step": 3036
    },
    {
      "epoch": 0.2189144381172061,
      "grad_norm": 0.22492740431580419,
      "learning_rate": 0.0001818675624300685,
      "loss": 1.0959,
      "step": 3037
    },
    {
      "epoch": 0.218986520579543,
      "grad_norm": 1.761405174417724,
      "learning_rate": 0.0001818541529518633,
      "loss": 1.1438,
      "step": 3038
    },
    {
      "epoch": 0.2190586030418799,
      "grad_norm": 0.14232818023363444,
      "learning_rate": 0.0001818407390118744,
      "loss": 1.0693,
      "step": 3039
    },
    {
      "epoch": 0.21913068550421683,
      "grad_norm": 0.29579617312240025,
      "learning_rate": 0.0001818273206108331,
      "loss": 1.0756,
      "step": 3040
    },
    {
      "epoch": 0.21920276796655375,
      "grad_norm": 0.18804384201731972,
      "learning_rate": 0.00018181389774947078,
      "loss": 1.0406,
      "step": 3041
    },
    {
      "epoch": 0.21927485042889064,
      "grad_norm": 0.4906415096852782,
      "learning_rate": 0.0001818004704285191,
      "loss": 1.2704,
      "step": 3042
    },
    {
      "epoch": 0.21934693289122756,
      "grad_norm": 0.8990407071715534,
      "learning_rate": 0.00018178703864870998,
      "loss": 1.0324,
      "step": 3043
    },
    {
      "epoch": 0.21941901535356448,
      "grad_norm": 0.13921228759179555,
      "learning_rate": 0.00018177360241077557,
      "loss": 1.0479,
      "step": 3044
    },
    {
      "epoch": 0.2194910978159014,
      "grad_norm": 0.28195932376204774,
      "learning_rate": 0.00018176016171544826,
      "loss": 1.1371,
      "step": 3045
    },
    {
      "epoch": 0.21956318027823832,
      "grad_norm": 0.20427678113232517,
      "learning_rate": 0.00018174671656346069,
      "loss": 1.0149,
      "step": 3046
    },
    {
      "epoch": 0.2196352627405752,
      "grad_norm": 0.23421815934759624,
      "learning_rate": 0.0001817332669555457,
      "loss": 1.0707,
      "step": 3047
    },
    {
      "epoch": 0.21970734520291213,
      "grad_norm": 0.22376308095449315,
      "learning_rate": 0.0001817198128924365,
      "loss": 1.0427,
      "step": 3048
    },
    {
      "epoch": 0.21977942766524905,
      "grad_norm": 0.1383096491059263,
      "learning_rate": 0.0001817063543748664,
      "loss": 1.0683,
      "step": 3049
    },
    {
      "epoch": 0.21985151012758597,
      "grad_norm": 0.17570633667175903,
      "learning_rate": 0.00018169289140356902,
      "loss": 1.1101,
      "step": 3050
    },
    {
      "epoch": 0.21992359258992286,
      "grad_norm": 0.33400583505719905,
      "learning_rate": 0.0001816794239792782,
      "loss": 1.1148,
      "step": 3051
    },
    {
      "epoch": 0.21999567505225978,
      "grad_norm": 0.22456646561812169,
      "learning_rate": 0.00018166595210272805,
      "loss": 1.0987,
      "step": 3052
    },
    {
      "epoch": 0.2200677575145967,
      "grad_norm": 0.2022245362921945,
      "learning_rate": 0.00018165247577465294,
      "loss": 1.0791,
      "step": 3053
    },
    {
      "epoch": 0.22013983997693362,
      "grad_norm": 0.16087844457079148,
      "learning_rate": 0.0001816389949957874,
      "loss": 1.1534,
      "step": 3054
    },
    {
      "epoch": 0.2202119224392705,
      "grad_norm": 0.18964954599387682,
      "learning_rate": 0.00018162550976686626,
      "loss": 1.1149,
      "step": 3055
    },
    {
      "epoch": 0.22028400490160743,
      "grad_norm": 0.13404213684877045,
      "learning_rate": 0.00018161202008862458,
      "loss": 1.0321,
      "step": 3056
    },
    {
      "epoch": 0.22035608736394435,
      "grad_norm": 0.22264854135950538,
      "learning_rate": 0.0001815985259617977,
      "loss": 1.1475,
      "step": 3057
    },
    {
      "epoch": 0.22042816982628127,
      "grad_norm": 0.1383552449467262,
      "learning_rate": 0.00018158502738712114,
      "loss": 1.1299,
      "step": 3058
    },
    {
      "epoch": 0.2205002522886182,
      "grad_norm": 1.6217602814035668,
      "learning_rate": 0.00018157152436533072,
      "loss": 1.1043,
      "step": 3059
    },
    {
      "epoch": 0.22057233475095508,
      "grad_norm": 0.1595276528887851,
      "learning_rate": 0.00018155801689716245,
      "loss": 1.0512,
      "step": 3060
    },
    {
      "epoch": 0.220644417213292,
      "grad_norm": 0.21358213735953593,
      "learning_rate": 0.00018154450498335264,
      "loss": 1.0544,
      "step": 3061
    },
    {
      "epoch": 0.22071649967562892,
      "grad_norm": 0.21566227547524258,
      "learning_rate": 0.00018153098862463778,
      "loss": 1.1073,
      "step": 3062
    },
    {
      "epoch": 0.22078858213796584,
      "grad_norm": 0.13300156002328695,
      "learning_rate": 0.00018151746782175466,
      "loss": 1.0994,
      "step": 3063
    },
    {
      "epoch": 0.22086066460030274,
      "grad_norm": 0.16789173801197604,
      "learning_rate": 0.00018150394257544027,
      "loss": 1.1313,
      "step": 3064
    },
    {
      "epoch": 0.22093274706263966,
      "grad_norm": 0.408377832309459,
      "learning_rate": 0.00018149041288643183,
      "loss": 1.2316,
      "step": 3065
    },
    {
      "epoch": 0.22100482952497658,
      "grad_norm": 0.5500749193741222,
      "learning_rate": 0.00018147687875546685,
      "loss": 1.0868,
      "step": 3066
    },
    {
      "epoch": 0.2210769119873135,
      "grad_norm": 0.313883469613294,
      "learning_rate": 0.00018146334018328308,
      "loss": 1.1271,
      "step": 3067
    },
    {
      "epoch": 0.2211489944496504,
      "grad_norm": 0.20427122143796828,
      "learning_rate": 0.00018144979717061847,
      "loss": 1.1105,
      "step": 3068
    },
    {
      "epoch": 0.2212210769119873,
      "grad_norm": 0.21581157041901577,
      "learning_rate": 0.00018143624971821124,
      "loss": 1.1013,
      "step": 3069
    },
    {
      "epoch": 0.22129315937432423,
      "grad_norm": 0.14951589180560473,
      "learning_rate": 0.00018142269782679988,
      "loss": 1.0429,
      "step": 3070
    },
    {
      "epoch": 0.22136524183666115,
      "grad_norm": 0.14769699189714167,
      "learning_rate": 0.00018140914149712302,
      "loss": 1.0713,
      "step": 3071
    },
    {
      "epoch": 0.22143732429899807,
      "grad_norm": 0.25028853059393863,
      "learning_rate": 0.00018139558072991966,
      "loss": 1.1824,
      "step": 3072
    },
    {
      "epoch": 0.22150940676133496,
      "grad_norm": 0.23497640862394673,
      "learning_rate": 0.00018138201552592896,
      "loss": 1.1078,
      "step": 3073
    },
    {
      "epoch": 0.22158148922367188,
      "grad_norm": 0.12157076406123767,
      "learning_rate": 0.00018136844588589036,
      "loss": 1.0665,
      "step": 3074
    },
    {
      "epoch": 0.2216535716860088,
      "grad_norm": 0.5025012976029523,
      "learning_rate": 0.00018135487181054351,
      "loss": 1.0607,
      "step": 3075
    },
    {
      "epoch": 0.22172565414834572,
      "grad_norm": 0.13890987355595094,
      "learning_rate": 0.00018134129330062834,
      "loss": 1.0581,
      "step": 3076
    },
    {
      "epoch": 0.2217977366106826,
      "grad_norm": 0.21399246622011123,
      "learning_rate": 0.00018132771035688498,
      "loss": 1.1351,
      "step": 3077
    },
    {
      "epoch": 0.22186981907301953,
      "grad_norm": 0.17480512012899593,
      "learning_rate": 0.0001813141229800538,
      "loss": 1.1067,
      "step": 3078
    },
    {
      "epoch": 0.22194190153535645,
      "grad_norm": 0.34097746274538476,
      "learning_rate": 0.00018130053117087547,
      "loss": 1.064,
      "step": 3079
    },
    {
      "epoch": 0.22201398399769337,
      "grad_norm": 0.1730355960706431,
      "learning_rate": 0.00018128693493009088,
      "loss": 1.0951,
      "step": 3080
    },
    {
      "epoch": 0.22208606646003026,
      "grad_norm": 0.17850592435058735,
      "learning_rate": 0.00018127333425844107,
      "loss": 1.0661,
      "step": 3081
    },
    {
      "epoch": 0.22215814892236718,
      "grad_norm": 0.19948165059369566,
      "learning_rate": 0.00018125972915666752,
      "loss": 1.16,
      "step": 3082
    },
    {
      "epoch": 0.2222302313847041,
      "grad_norm": 0.17009609271683176,
      "learning_rate": 0.00018124611962551173,
      "loss": 0.9975,
      "step": 3083
    },
    {
      "epoch": 0.22230231384704102,
      "grad_norm": 0.1159791416014358,
      "learning_rate": 0.00018123250566571557,
      "loss": 1.0913,
      "step": 3084
    },
    {
      "epoch": 0.22237439630937794,
      "grad_norm": 0.14986916573786854,
      "learning_rate": 0.00018121888727802113,
      "loss": 1.1608,
      "step": 3085
    },
    {
      "epoch": 0.22244647877171483,
      "grad_norm": 0.15549843630514776,
      "learning_rate": 0.00018120526446317072,
      "loss": 1.1166,
      "step": 3086
    },
    {
      "epoch": 0.22251856123405175,
      "grad_norm": 0.13389984924883513,
      "learning_rate": 0.0001811916372219069,
      "loss": 1.0336,
      "step": 3087
    },
    {
      "epoch": 0.22259064369638867,
      "grad_norm": 0.22355151704243326,
      "learning_rate": 0.00018117800555497252,
      "loss": 1.0394,
      "step": 3088
    },
    {
      "epoch": 0.2226627261587256,
      "grad_norm": 0.16207476909204258,
      "learning_rate": 0.00018116436946311057,
      "loss": 1.0746,
      "step": 3089
    },
    {
      "epoch": 0.22273480862106249,
      "grad_norm": 0.12666886885787546,
      "learning_rate": 0.0001811507289470644,
      "loss": 1.0613,
      "step": 3090
    },
    {
      "epoch": 0.2228068910833994,
      "grad_norm": 1.4237347415239263,
      "learning_rate": 0.0001811370840075775,
      "loss": 1.0511,
      "step": 3091
    },
    {
      "epoch": 0.22287897354573633,
      "grad_norm": 0.18577192043501076,
      "learning_rate": 0.00018112343464539362,
      "loss": 1.1487,
      "step": 3092
    },
    {
      "epoch": 0.22295105600807325,
      "grad_norm": 0.21169121333306717,
      "learning_rate": 0.00018110978086125683,
      "loss": 1.1459,
      "step": 3093
    },
    {
      "epoch": 0.22302313847041014,
      "grad_norm": 0.39189522701224927,
      "learning_rate": 0.00018109612265591134,
      "loss": 1.1102,
      "step": 3094
    },
    {
      "epoch": 0.22309522093274706,
      "grad_norm": 0.27759054727951865,
      "learning_rate": 0.00018108246003010166,
      "loss": 1.0725,
      "step": 3095
    },
    {
      "epoch": 0.22316730339508398,
      "grad_norm": 0.1467006992754567,
      "learning_rate": 0.0001810687929845725,
      "loss": 1.0245,
      "step": 3096
    },
    {
      "epoch": 0.2232393858574209,
      "grad_norm": 0.3212766804426659,
      "learning_rate": 0.00018105512152006892,
      "loss": 1.0704,
      "step": 3097
    },
    {
      "epoch": 0.22331146831975782,
      "grad_norm": 0.13121508462732112,
      "learning_rate": 0.00018104144563733604,
      "loss": 1.0557,
      "step": 3098
    },
    {
      "epoch": 0.2233835507820947,
      "grad_norm": 0.11488138238974044,
      "learning_rate": 0.00018102776533711936,
      "loss": 1.1419,
      "step": 3099
    },
    {
      "epoch": 0.22345563324443163,
      "grad_norm": 0.18678477422070464,
      "learning_rate": 0.00018101408062016455,
      "loss": 1.0529,
      "step": 3100
    },
    {
      "epoch": 0.22352771570676855,
      "grad_norm": 0.276218711491562,
      "learning_rate": 0.00018100039148721763,
      "loss": 1.0962,
      "step": 3101
    },
    {
      "epoch": 0.22359979816910547,
      "grad_norm": 0.2088722764222868,
      "learning_rate": 0.00018098669793902467,
      "loss": 1.1419,
      "step": 3102
    },
    {
      "epoch": 0.22367188063144236,
      "grad_norm": 0.14233640669717337,
      "learning_rate": 0.00018097299997633218,
      "loss": 1.083,
      "step": 3103
    },
    {
      "epoch": 0.22374396309377928,
      "grad_norm": 0.16753196349581725,
      "learning_rate": 0.00018095929759988675,
      "loss": 1.1144,
      "step": 3104
    },
    {
      "epoch": 0.2238160455561162,
      "grad_norm": 0.16793970400761496,
      "learning_rate": 0.00018094559081043533,
      "loss": 1.0744,
      "step": 3105
    },
    {
      "epoch": 0.22388812801845312,
      "grad_norm": 0.22648777193567726,
      "learning_rate": 0.00018093187960872506,
      "loss": 1.1557,
      "step": 3106
    },
    {
      "epoch": 0.22396021048079,
      "grad_norm": 0.4506102085314626,
      "learning_rate": 0.0001809181639955033,
      "loss": 1.0731,
      "step": 3107
    },
    {
      "epoch": 0.22403229294312693,
      "grad_norm": 0.14501907636664088,
      "learning_rate": 0.00018090444397151774,
      "loss": 1.0575,
      "step": 3108
    },
    {
      "epoch": 0.22410437540546385,
      "grad_norm": 0.1738618292984959,
      "learning_rate": 0.00018089071953751614,
      "loss": 1.0608,
      "step": 3109
    },
    {
      "epoch": 0.22417645786780077,
      "grad_norm": 0.25424771654948086,
      "learning_rate": 0.00018087699069424667,
      "loss": 1.1152,
      "step": 3110
    },
    {
      "epoch": 0.2242485403301377,
      "grad_norm": 0.632765076564337,
      "learning_rate": 0.00018086325744245767,
      "loss": 1.0523,
      "step": 3111
    },
    {
      "epoch": 0.22432062279247458,
      "grad_norm": 0.13118627062618893,
      "learning_rate": 0.0001808495197828977,
      "loss": 1.0749,
      "step": 3112
    },
    {
      "epoch": 0.2243927052548115,
      "grad_norm": 0.29672664521732856,
      "learning_rate": 0.00018083577771631562,
      "loss": 1.0496,
      "step": 3113
    },
    {
      "epoch": 0.22446478771714842,
      "grad_norm": 0.1774192688485085,
      "learning_rate": 0.00018082203124346045,
      "loss": 1.1173,
      "step": 3114
    },
    {
      "epoch": 0.22453687017948534,
      "grad_norm": 0.1817250863117363,
      "learning_rate": 0.00018080828036508154,
      "loss": 1.1147,
      "step": 3115
    },
    {
      "epoch": 0.22460895264182223,
      "grad_norm": 0.11316184034374374,
      "learning_rate": 0.00018079452508192842,
      "loss": 1.0669,
      "step": 3116
    },
    {
      "epoch": 0.22468103510415915,
      "grad_norm": 0.18575697522409718,
      "learning_rate": 0.00018078076539475085,
      "loss": 1.1432,
      "step": 3117
    },
    {
      "epoch": 0.22475311756649607,
      "grad_norm": 0.2260978335738606,
      "learning_rate": 0.0001807670013042989,
      "loss": 1.0028,
      "step": 3118
    },
    {
      "epoch": 0.224825200028833,
      "grad_norm": 0.12517210150817576,
      "learning_rate": 0.00018075323281132284,
      "loss": 1.0911,
      "step": 3119
    },
    {
      "epoch": 0.2248972824911699,
      "grad_norm": 0.13710820150895672,
      "learning_rate": 0.00018073945991657313,
      "loss": 1.0763,
      "step": 3120
    },
    {
      "epoch": 0.2249693649535068,
      "grad_norm": 0.12882546312612306,
      "learning_rate": 0.00018072568262080053,
      "loss": 1.0778,
      "step": 3121
    },
    {
      "epoch": 0.22504144741584373,
      "grad_norm": 0.10690258741010003,
      "learning_rate": 0.00018071190092475603,
      "loss": 0.9654,
      "step": 3122
    },
    {
      "epoch": 0.22511352987818065,
      "grad_norm": 0.2604039630973547,
      "learning_rate": 0.0001806981148291909,
      "loss": 1.1325,
      "step": 3123
    },
    {
      "epoch": 0.22518561234051757,
      "grad_norm": 0.11804020068453926,
      "learning_rate": 0.00018068432433485652,
      "loss": 1.0562,
      "step": 3124
    },
    {
      "epoch": 0.22525769480285446,
      "grad_norm": 0.1959006347550568,
      "learning_rate": 0.00018067052944250467,
      "loss": 1.158,
      "step": 3125
    },
    {
      "epoch": 0.22532977726519138,
      "grad_norm": 0.15477642829523863,
      "learning_rate": 0.00018065673015288725,
      "loss": 1.0592,
      "step": 3126
    },
    {
      "epoch": 0.2254018597275283,
      "grad_norm": 0.1408211201243355,
      "learning_rate": 0.00018064292646675646,
      "loss": 1.1079,
      "step": 3127
    },
    {
      "epoch": 0.22547394218986522,
      "grad_norm": 0.13526808803457133,
      "learning_rate": 0.00018062911838486473,
      "loss": 1.0603,
      "step": 3128
    },
    {
      "epoch": 0.2255460246522021,
      "grad_norm": 0.20229402668290902,
      "learning_rate": 0.00018061530590796475,
      "loss": 1.1592,
      "step": 3129
    },
    {
      "epoch": 0.22561810711453903,
      "grad_norm": 0.16320464947592891,
      "learning_rate": 0.00018060148903680933,
      "loss": 1.165,
      "step": 3130
    },
    {
      "epoch": 0.22569018957687595,
      "grad_norm": 0.2285817334274781,
      "learning_rate": 0.0001805876677721517,
      "loss": 1.0034,
      "step": 3131
    },
    {
      "epoch": 0.22576227203921287,
      "grad_norm": 0.20076773593991293,
      "learning_rate": 0.00018057384211474527,
      "loss": 1.15,
      "step": 3132
    },
    {
      "epoch": 0.22583435450154976,
      "grad_norm": 0.18171704286570622,
      "learning_rate": 0.00018056001206534356,
      "loss": 1.0619,
      "step": 3133
    },
    {
      "epoch": 0.22590643696388668,
      "grad_norm": 0.15500595683612434,
      "learning_rate": 0.0001805461776247005,
      "loss": 1.046,
      "step": 3134
    },
    {
      "epoch": 0.2259785194262236,
      "grad_norm": 0.13374125959972444,
      "learning_rate": 0.00018053233879357014,
      "loss": 1.0274,
      "step": 3135
    },
    {
      "epoch": 0.22605060188856052,
      "grad_norm": 0.15175208022102582,
      "learning_rate": 0.00018051849557270688,
      "loss": 1.0051,
      "step": 3136
    },
    {
      "epoch": 0.22612268435089744,
      "grad_norm": 0.21250658529853284,
      "learning_rate": 0.00018050464796286528,
      "loss": 1.1075,
      "step": 3137
    },
    {
      "epoch": 0.22619476681323433,
      "grad_norm": 0.22661908048207613,
      "learning_rate": 0.00018049079596480015,
      "loss": 1.0861,
      "step": 3138
    },
    {
      "epoch": 0.22626684927557125,
      "grad_norm": 0.19493210847143147,
      "learning_rate": 0.00018047693957926652,
      "loss": 1.1059,
      "step": 3139
    },
    {
      "epoch": 0.22633893173790817,
      "grad_norm": 0.1385092961988166,
      "learning_rate": 0.00018046307880701976,
      "loss": 1.0959,
      "step": 3140
    },
    {
      "epoch": 0.2264110142002451,
      "grad_norm": 0.14159952816021515,
      "learning_rate": 0.00018044921364881533,
      "loss": 1.077,
      "step": 3141
    },
    {
      "epoch": 0.22648309666258198,
      "grad_norm": 0.1495153730907291,
      "learning_rate": 0.00018043534410540905,
      "loss": 1.0711,
      "step": 3142
    },
    {
      "epoch": 0.2265551791249189,
      "grad_norm": 0.15744883970850618,
      "learning_rate": 0.0001804214701775569,
      "loss": 1.0777,
      "step": 3143
    },
    {
      "epoch": 0.22662726158725582,
      "grad_norm": 0.1521656892990839,
      "learning_rate": 0.00018040759186601518,
      "loss": 1.0837,
      "step": 3144
    },
    {
      "epoch": 0.22669934404959274,
      "grad_norm": 0.17232947983145233,
      "learning_rate": 0.00018039370917154036,
      "loss": 1.1202,
      "step": 3145
    },
    {
      "epoch": 0.22677142651192964,
      "grad_norm": 0.14149518790449064,
      "learning_rate": 0.00018037982209488914,
      "loss": 1.0223,
      "step": 3146
    },
    {
      "epoch": 0.22684350897426656,
      "grad_norm": 0.13175299567360108,
      "learning_rate": 0.00018036593063681852,
      "loss": 1.076,
      "step": 3147
    },
    {
      "epoch": 0.22691559143660348,
      "grad_norm": 0.1432868062014197,
      "learning_rate": 0.00018035203479808574,
      "loss": 1.1464,
      "step": 3148
    },
    {
      "epoch": 0.2269876738989404,
      "grad_norm": 0.14016561628637145,
      "learning_rate": 0.00018033813457944816,
      "loss": 1.1637,
      "step": 3149
    },
    {
      "epoch": 0.22705975636127732,
      "grad_norm": 0.13256262203799504,
      "learning_rate": 0.00018032422998166356,
      "loss": 1.0938,
      "step": 3150
    },
    {
      "epoch": 0.2271318388236142,
      "grad_norm": 0.15980024530397274,
      "learning_rate": 0.0001803103210054898,
      "loss": 1.0744,
      "step": 3151
    },
    {
      "epoch": 0.22720392128595113,
      "grad_norm": 0.983907784589095,
      "learning_rate": 0.0001802964076516851,
      "loss": 1.186,
      "step": 3152
    },
    {
      "epoch": 0.22727600374828805,
      "grad_norm": 0.1336506468253119,
      "learning_rate": 0.00018028248992100782,
      "loss": 1.0994,
      "step": 3153
    },
    {
      "epoch": 0.22734808621062497,
      "grad_norm": 1.4823852402746738,
      "learning_rate": 0.0001802685678142166,
      "loss": 1.1406,
      "step": 3154
    },
    {
      "epoch": 0.22742016867296186,
      "grad_norm": 0.3516299081712621,
      "learning_rate": 0.00018025464133207035,
      "loss": 1.1977,
      "step": 3155
    },
    {
      "epoch": 0.22749225113529878,
      "grad_norm": 0.16011594752683214,
      "learning_rate": 0.00018024071047532817,
      "loss": 0.9585,
      "step": 3156
    },
    {
      "epoch": 0.2275643335976357,
      "grad_norm": 0.15493808265300785,
      "learning_rate": 0.00018022677524474938,
      "loss": 1.0412,
      "step": 3157
    },
    {
      "epoch": 0.22763641605997262,
      "grad_norm": 0.13439084796504347,
      "learning_rate": 0.00018021283564109364,
      "loss": 1.0737,
      "step": 3158
    },
    {
      "epoch": 0.2277084985223095,
      "grad_norm": 0.1557892303620582,
      "learning_rate": 0.00018019889166512074,
      "loss": 1.145,
      "step": 3159
    },
    {
      "epoch": 0.22778058098464643,
      "grad_norm": 0.31464877094370314,
      "learning_rate": 0.00018018494331759077,
      "loss": 1.0212,
      "step": 3160
    },
    {
      "epoch": 0.22785266344698335,
      "grad_norm": 0.2256876750587185,
      "learning_rate": 0.00018017099059926403,
      "loss": 1.0889,
      "step": 3161
    },
    {
      "epoch": 0.22792474590932027,
      "grad_norm": 0.16797166430985436,
      "learning_rate": 0.00018015703351090107,
      "loss": 1.0685,
      "step": 3162
    },
    {
      "epoch": 0.22799682837165716,
      "grad_norm": 0.1651116439446229,
      "learning_rate": 0.00018014307205326268,
      "loss": 1.0162,
      "step": 3163
    },
    {
      "epoch": 0.22806891083399408,
      "grad_norm": 0.12087081766024257,
      "learning_rate": 0.0001801291062271099,
      "loss": 1.1215,
      "step": 3164
    },
    {
      "epoch": 0.228140993296331,
      "grad_norm": 0.11713007578243581,
      "learning_rate": 0.00018011513603320397,
      "loss": 1.0765,
      "step": 3165
    },
    {
      "epoch": 0.22821307575866792,
      "grad_norm": 0.39987103696648413,
      "learning_rate": 0.00018010116147230637,
      "loss": 1.0571,
      "step": 3166
    },
    {
      "epoch": 0.22828515822100484,
      "grad_norm": 0.11181314248396163,
      "learning_rate": 0.0001800871825451789,
      "loss": 1.0819,
      "step": 3167
    },
    {
      "epoch": 0.22835724068334173,
      "grad_norm": 0.3684354880877232,
      "learning_rate": 0.00018007319925258348,
      "loss": 1.0357,
      "step": 3168
    },
    {
      "epoch": 0.22842932314567865,
      "grad_norm": 0.6277421492953029,
      "learning_rate": 0.00018005921159528233,
      "loss": 1.1144,
      "step": 3169
    },
    {
      "epoch": 0.22850140560801557,
      "grad_norm": 0.13213867002621876,
      "learning_rate": 0.00018004521957403795,
      "loss": 1.0885,
      "step": 3170
    },
    {
      "epoch": 0.2285734880703525,
      "grad_norm": 0.28727479744404566,
      "learning_rate": 0.000180031223189613,
      "loss": 1.0488,
      "step": 3171
    },
    {
      "epoch": 0.22864557053268939,
      "grad_norm": 0.11370115597449998,
      "learning_rate": 0.00018001722244277035,
      "loss": 1.0612,
      "step": 3172
    },
    {
      "epoch": 0.2287176529950263,
      "grad_norm": 0.12387007027531582,
      "learning_rate": 0.00018000321733427328,
      "loss": 1.0292,
      "step": 3173
    },
    {
      "epoch": 0.22878973545736322,
      "grad_norm": 0.39411434022752767,
      "learning_rate": 0.0001799892078648851,
      "loss": 1.1515,
      "step": 3174
    },
    {
      "epoch": 0.22886181791970014,
      "grad_norm": 0.1775354337622811,
      "learning_rate": 0.00017997519403536953,
      "loss": 0.987,
      "step": 3175
    },
    {
      "epoch": 0.22893390038203704,
      "grad_norm": 0.4978844299127917,
      "learning_rate": 0.00017996117584649038,
      "loss": 1.1574,
      "step": 3176
    },
    {
      "epoch": 0.22900598284437396,
      "grad_norm": 1.974090525941313,
      "learning_rate": 0.0001799471532990118,
      "loss": 1.089,
      "step": 3177
    },
    {
      "epoch": 0.22907806530671088,
      "grad_norm": 0.44687296307743946,
      "learning_rate": 0.00017993312639369812,
      "loss": 1.1068,
      "step": 3178
    },
    {
      "epoch": 0.2291501477690478,
      "grad_norm": 0.28722252375174145,
      "learning_rate": 0.00017991909513131397,
      "loss": 1.0398,
      "step": 3179
    },
    {
      "epoch": 0.22922223023138472,
      "grad_norm": 0.16471176262481949,
      "learning_rate": 0.00017990505951262416,
      "loss": 1.0509,
      "step": 3180
    },
    {
      "epoch": 0.2292943126937216,
      "grad_norm": 0.32762900678223517,
      "learning_rate": 0.00017989101953839375,
      "loss": 1.1988,
      "step": 3181
    },
    {
      "epoch": 0.22936639515605853,
      "grad_norm": 0.22526170954040098,
      "learning_rate": 0.00017987697520938803,
      "loss": 1.095,
      "step": 3182
    },
    {
      "epoch": 0.22943847761839545,
      "grad_norm": 0.16263975495290728,
      "learning_rate": 0.00017986292652637259,
      "loss": 1.1519,
      "step": 3183
    },
    {
      "epoch": 0.22951056008073237,
      "grad_norm": 0.2800137173162009,
      "learning_rate": 0.00017984887349011314,
      "loss": 1.0586,
      "step": 3184
    },
    {
      "epoch": 0.22958264254306926,
      "grad_norm": 0.13799417505077502,
      "learning_rate": 0.00017983481610137577,
      "loss": 1.1646,
      "step": 3185
    },
    {
      "epoch": 0.22965472500540618,
      "grad_norm": 0.15849432070837227,
      "learning_rate": 0.00017982075436092672,
      "loss": 1.0089,
      "step": 3186
    },
    {
      "epoch": 0.2297268074677431,
      "grad_norm": 0.2259699519204055,
      "learning_rate": 0.00017980668826953245,
      "loss": 1.0657,
      "step": 3187
    },
    {
      "epoch": 0.22979888993008002,
      "grad_norm": 0.12814752751272684,
      "learning_rate": 0.00017979261782795965,
      "loss": 1.0151,
      "step": 3188
    },
    {
      "epoch": 0.2298709723924169,
      "grad_norm": 0.15998650340062434,
      "learning_rate": 0.0001797785430369754,
      "loss": 1.0805,
      "step": 3189
    },
    {
      "epoch": 0.22994305485475383,
      "grad_norm": 0.17502688069802194,
      "learning_rate": 0.00017976446389734676,
      "loss": 1.0138,
      "step": 3190
    },
    {
      "epoch": 0.23001513731709075,
      "grad_norm": 0.6315613543265358,
      "learning_rate": 0.0001797503804098413,
      "loss": 1.101,
      "step": 3191
    },
    {
      "epoch": 0.23008721977942767,
      "grad_norm": 0.1359335334308489,
      "learning_rate": 0.0001797362925752266,
      "loss": 1.0279,
      "step": 3192
    },
    {
      "epoch": 0.2301593022417646,
      "grad_norm": 0.11948872143911049,
      "learning_rate": 0.00017972220039427067,
      "loss": 0.9637,
      "step": 3193
    },
    {
      "epoch": 0.23023138470410148,
      "grad_norm": 0.15131928144559686,
      "learning_rate": 0.00017970810386774155,
      "loss": 1.0802,
      "step": 3194
    },
    {
      "epoch": 0.2303034671664384,
      "grad_norm": 0.18481166434942126,
      "learning_rate": 0.00017969400299640772,
      "loss": 1.1133,
      "step": 3195
    },
    {
      "epoch": 0.23037554962877532,
      "grad_norm": 0.1556950750942223,
      "learning_rate": 0.00017967989778103774,
      "loss": 1.0798,
      "step": 3196
    },
    {
      "epoch": 0.23044763209111224,
      "grad_norm": 0.239402463047565,
      "learning_rate": 0.0001796657882224005,
      "loss": 1.0783,
      "step": 3197
    },
    {
      "epoch": 0.23051971455344913,
      "grad_norm": 1.3786920271187788,
      "learning_rate": 0.0001796516743212651,
      "loss": 1.0776,
      "step": 3198
    },
    {
      "epoch": 0.23059179701578605,
      "grad_norm": 0.20658706853598122,
      "learning_rate": 0.00017963755607840084,
      "loss": 1.1366,
      "step": 3199
    },
    {
      "epoch": 0.23066387947812297,
      "grad_norm": 0.9247168197233836,
      "learning_rate": 0.00017962343349457735,
      "loss": 1.2161,
      "step": 3200
    },
    {
      "epoch": 0.2307359619404599,
      "grad_norm": 0.13218805654833135,
      "learning_rate": 0.00017960930657056438,
      "loss": 1.05,
      "step": 3201
    },
    {
      "epoch": 0.23080804440279679,
      "grad_norm": 0.42668777619329207,
      "learning_rate": 0.00017959517530713202,
      "loss": 1.1116,
      "step": 3202
    },
    {
      "epoch": 0.2308801268651337,
      "grad_norm": 0.15546144168180537,
      "learning_rate": 0.0001795810397050505,
      "loss": 1.0527,
      "step": 3203
    },
    {
      "epoch": 0.23095220932747063,
      "grad_norm": 0.1994086026701391,
      "learning_rate": 0.00017956689976509038,
      "loss": 1.1257,
      "step": 3204
    },
    {
      "epoch": 0.23102429178980755,
      "grad_norm": 0.17586992551008568,
      "learning_rate": 0.0001795527554880224,
      "loss": 1.0275,
      "step": 3205
    },
    {
      "epoch": 0.23109637425214447,
      "grad_norm": 0.19387119861220117,
      "learning_rate": 0.0001795386068746175,
      "loss": 1.0276,
      "step": 3206
    },
    {
      "epoch": 0.23116845671448136,
      "grad_norm": 0.4136309889894153,
      "learning_rate": 0.00017952445392564701,
      "loss": 1.0945,
      "step": 3207
    },
    {
      "epoch": 0.23124053917681828,
      "grad_norm": 0.1525111262736152,
      "learning_rate": 0.00017951029664188233,
      "loss": 1.0831,
      "step": 3208
    },
    {
      "epoch": 0.2313126216391552,
      "grad_norm": 0.2905735643726876,
      "learning_rate": 0.0001794961350240951,
      "loss": 1.0576,
      "step": 3209
    },
    {
      "epoch": 0.23138470410149212,
      "grad_norm": 0.26862040277191945,
      "learning_rate": 0.00017948196907305742,
      "loss": 1.0609,
      "step": 3210
    },
    {
      "epoch": 0.231456786563829,
      "grad_norm": 0.13589778551349005,
      "learning_rate": 0.0001794677987895413,
      "loss": 1.1244,
      "step": 3211
    },
    {
      "epoch": 0.23152886902616593,
      "grad_norm": 0.635215130962783,
      "learning_rate": 0.00017945362417431918,
      "loss": 1.0393,
      "step": 3212
    },
    {
      "epoch": 0.23160095148850285,
      "grad_norm": 0.16723017808187318,
      "learning_rate": 0.00017943944522816379,
      "loss": 1.0113,
      "step": 3213
    },
    {
      "epoch": 0.23167303395083977,
      "grad_norm": 0.3580877912991005,
      "learning_rate": 0.0001794252619518479,
      "loss": 1.1341,
      "step": 3214
    },
    {
      "epoch": 0.23174511641317666,
      "grad_norm": 0.21459292916940192,
      "learning_rate": 0.0001794110743461447,
      "loss": 1.0467,
      "step": 3215
    },
    {
      "epoch": 0.23181719887551358,
      "grad_norm": 0.1513793799413487,
      "learning_rate": 0.00017939688241182751,
      "loss": 1.0065,
      "step": 3216
    },
    {
      "epoch": 0.2318892813378505,
      "grad_norm": 0.13484619337059167,
      "learning_rate": 0.00017938268614966994,
      "loss": 1.0939,
      "step": 3217
    },
    {
      "epoch": 0.23196136380018742,
      "grad_norm": 0.21816490213641945,
      "learning_rate": 0.00017936848556044578,
      "loss": 1.0803,
      "step": 3218
    },
    {
      "epoch": 0.23203344626252434,
      "grad_norm": 0.13743867525967518,
      "learning_rate": 0.00017935428064492912,
      "loss": 1.0974,
      "step": 3219
    },
    {
      "epoch": 0.23210552872486123,
      "grad_norm": 0.24352855014198488,
      "learning_rate": 0.00017934007140389425,
      "loss": 1.1064,
      "step": 3220
    },
    {
      "epoch": 0.23217761118719815,
      "grad_norm": 0.1362776955596667,
      "learning_rate": 0.00017932585783811565,
      "loss": 1.0318,
      "step": 3221
    },
    {
      "epoch": 0.23224969364953507,
      "grad_norm": 0.20006272494048352,
      "learning_rate": 0.00017931163994836817,
      "loss": 1.1109,
      "step": 3222
    },
    {
      "epoch": 0.232321776111872,
      "grad_norm": 0.2252034386052371,
      "learning_rate": 0.00017929741773542674,
      "loss": 1.1229,
      "step": 3223
    },
    {
      "epoch": 0.23239385857420888,
      "grad_norm": 0.14469927237743144,
      "learning_rate": 0.00017928319120006663,
      "loss": 1.014,
      "step": 3224
    },
    {
      "epoch": 0.2324659410365458,
      "grad_norm": 0.6126213510102125,
      "learning_rate": 0.00017926896034306332,
      "loss": 1.1406,
      "step": 3225
    },
    {
      "epoch": 0.23253802349888272,
      "grad_norm": 0.11644290737882651,
      "learning_rate": 0.00017925472516519252,
      "loss": 1.0929,
      "step": 3226
    },
    {
      "epoch": 0.23261010596121964,
      "grad_norm": 0.42124966740317404,
      "learning_rate": 0.00017924048566723014,
      "loss": 1.0571,
      "step": 3227
    },
    {
      "epoch": 0.23268218842355654,
      "grad_norm": 0.14635587773708292,
      "learning_rate": 0.0001792262418499524,
      "loss": 1.085,
      "step": 3228
    },
    {
      "epoch": 0.23275427088589346,
      "grad_norm": 0.23449294706834495,
      "learning_rate": 0.00017921199371413575,
      "loss": 1.0638,
      "step": 3229
    },
    {
      "epoch": 0.23282635334823037,
      "grad_norm": 0.8328851499997244,
      "learning_rate": 0.00017919774126055673,
      "loss": 1.0864,
      "step": 3230
    },
    {
      "epoch": 0.2328984358105673,
      "grad_norm": 0.23979877534939958,
      "learning_rate": 0.0001791834844899923,
      "loss": 1.0751,
      "step": 3231
    },
    {
      "epoch": 0.23297051827290421,
      "grad_norm": 0.1701632808806383,
      "learning_rate": 0.00017916922340321958,
      "loss": 1.0879,
      "step": 3232
    },
    {
      "epoch": 0.2330426007352411,
      "grad_norm": 0.24866428553845427,
      "learning_rate": 0.00017915495800101594,
      "loss": 1.0295,
      "step": 3233
    },
    {
      "epoch": 0.23311468319757803,
      "grad_norm": 0.17654188796881978,
      "learning_rate": 0.0001791406882841589,
      "loss": 1.1622,
      "step": 3234
    },
    {
      "epoch": 0.23318676565991495,
      "grad_norm": 0.27428829430068324,
      "learning_rate": 0.00017912641425342637,
      "loss": 1.0461,
      "step": 3235
    },
    {
      "epoch": 0.23325884812225187,
      "grad_norm": 0.18594173424929722,
      "learning_rate": 0.00017911213590959636,
      "loss": 1.166,
      "step": 3236
    },
    {
      "epoch": 0.23333093058458876,
      "grad_norm": 0.13221072460846853,
      "learning_rate": 0.00017909785325344717,
      "loss": 1.1295,
      "step": 3237
    },
    {
      "epoch": 0.23340301304692568,
      "grad_norm": 0.11611692219717246,
      "learning_rate": 0.00017908356628575737,
      "loss": 1.0831,
      "step": 3238
    },
    {
      "epoch": 0.2334750955092626,
      "grad_norm": 0.15397041822694438,
      "learning_rate": 0.0001790692750073057,
      "loss": 1.1032,
      "step": 3239
    },
    {
      "epoch": 0.23354717797159952,
      "grad_norm": 0.20006361537445047,
      "learning_rate": 0.00017905497941887117,
      "loss": 1.0572,
      "step": 3240
    },
    {
      "epoch": 0.2336192604339364,
      "grad_norm": 0.23656726944005785,
      "learning_rate": 0.00017904067952123303,
      "loss": 1.1283,
      "step": 3241
    },
    {
      "epoch": 0.23369134289627333,
      "grad_norm": 0.33968263253432396,
      "learning_rate": 0.00017902637531517072,
      "loss": 1.0915,
      "step": 3242
    },
    {
      "epoch": 0.23376342535861025,
      "grad_norm": 0.17186189543621702,
      "learning_rate": 0.00017901206680146398,
      "loss": 1.0908,
      "step": 3243
    },
    {
      "epoch": 0.23383550782094717,
      "grad_norm": 0.16803520407653702,
      "learning_rate": 0.00017899775398089274,
      "loss": 1.1293,
      "step": 3244
    },
    {
      "epoch": 0.2339075902832841,
      "grad_norm": 0.12433402065164836,
      "learning_rate": 0.00017898343685423716,
      "loss": 1.0553,
      "step": 3245
    },
    {
      "epoch": 0.23397967274562098,
      "grad_norm": 0.11458340109074759,
      "learning_rate": 0.00017896911542227765,
      "loss": 1.113,
      "step": 3246
    },
    {
      "epoch": 0.2340517552079579,
      "grad_norm": 0.17309763168683093,
      "learning_rate": 0.0001789547896857949,
      "loss": 1.1071,
      "step": 3247
    },
    {
      "epoch": 0.23412383767029482,
      "grad_norm": 0.4087750447606269,
      "learning_rate": 0.00017894045964556975,
      "loss": 1.1694,
      "step": 3248
    },
    {
      "epoch": 0.23419592013263174,
      "grad_norm": 0.17284256086913724,
      "learning_rate": 0.00017892612530238334,
      "loss": 1.0907,
      "step": 3249
    },
    {
      "epoch": 0.23426800259496863,
      "grad_norm": 0.35982725789045267,
      "learning_rate": 0.00017891178665701697,
      "loss": 1.1426,
      "step": 3250
    },
    {
      "epoch": 0.23434008505730555,
      "grad_norm": 0.26391463396935966,
      "learning_rate": 0.0001788974437102523,
      "loss": 1.0232,
      "step": 3251
    },
    {
      "epoch": 0.23441216751964247,
      "grad_norm": 0.16560587980297195,
      "learning_rate": 0.00017888309646287113,
      "loss": 1.0348,
      "step": 3252
    },
    {
      "epoch": 0.2344842499819794,
      "grad_norm": 0.19230773705626789,
      "learning_rate": 0.00017886874491565544,
      "loss": 1.0823,
      "step": 3253
    },
    {
      "epoch": 0.23455633244431628,
      "grad_norm": 0.3313604653669703,
      "learning_rate": 0.0001788543890693876,
      "loss": 1.1332,
      "step": 3254
    },
    {
      "epoch": 0.2346284149066532,
      "grad_norm": 0.12788076160136827,
      "learning_rate": 0.0001788400289248501,
      "loss": 1.0908,
      "step": 3255
    },
    {
      "epoch": 0.23470049736899012,
      "grad_norm": 0.13977839973972175,
      "learning_rate": 0.0001788256644828257,
      "loss": 1.1061,
      "step": 3256
    },
    {
      "epoch": 0.23477257983132704,
      "grad_norm": 0.17105911240734353,
      "learning_rate": 0.0001788112957440974,
      "loss": 1.0775,
      "step": 3257
    },
    {
      "epoch": 0.23484466229366396,
      "grad_norm": 0.40452859543686415,
      "learning_rate": 0.0001787969227094484,
      "loss": 1.112,
      "step": 3258
    },
    {
      "epoch": 0.23491674475600086,
      "grad_norm": 0.5337384397709232,
      "learning_rate": 0.00017878254537966216,
      "loss": 1.1056,
      "step": 3259
    },
    {
      "epoch": 0.23498882721833778,
      "grad_norm": 0.2138370490775303,
      "learning_rate": 0.00017876816375552243,
      "loss": 1.096,
      "step": 3260
    },
    {
      "epoch": 0.2350609096806747,
      "grad_norm": 0.16614345280413406,
      "learning_rate": 0.0001787537778378131,
      "loss": 1.1418,
      "step": 3261
    },
    {
      "epoch": 0.23513299214301162,
      "grad_norm": 0.44786348369060625,
      "learning_rate": 0.0001787393876273183,
      "loss": 1.1058,
      "step": 3262
    },
    {
      "epoch": 0.2352050746053485,
      "grad_norm": 0.22953425068102457,
      "learning_rate": 0.00017872499312482246,
      "loss": 1.0706,
      "step": 3263
    },
    {
      "epoch": 0.23527715706768543,
      "grad_norm": 0.23790608263963123,
      "learning_rate": 0.00017871059433111018,
      "loss": 1.1575,
      "step": 3264
    },
    {
      "epoch": 0.23534923953002235,
      "grad_norm": 1.499244285427821,
      "learning_rate": 0.00017869619124696634,
      "loss": 1.0958,
      "step": 3265
    },
    {
      "epoch": 0.23542132199235927,
      "grad_norm": 0.20885862375874142,
      "learning_rate": 0.00017868178387317606,
      "loss": 1.0473,
      "step": 3266
    },
    {
      "epoch": 0.23549340445469616,
      "grad_norm": 0.12186638096371451,
      "learning_rate": 0.00017866737221052467,
      "loss": 1.158,
      "step": 3267
    },
    {
      "epoch": 0.23556548691703308,
      "grad_norm": 0.13042909108121092,
      "learning_rate": 0.0001786529562597977,
      "loss": 1.1528,
      "step": 3268
    },
    {
      "epoch": 0.23563756937937,
      "grad_norm": 0.22797368442541424,
      "learning_rate": 0.00017863853602178095,
      "loss": 0.9853,
      "step": 3269
    },
    {
      "epoch": 0.23570965184170692,
      "grad_norm": 0.7424289388812951,
      "learning_rate": 0.00017862411149726048,
      "loss": 1.2604,
      "step": 3270
    },
    {
      "epoch": 0.23578173430404384,
      "grad_norm": 0.31085423050638467,
      "learning_rate": 0.00017860968268702254,
      "loss": 1.157,
      "step": 3271
    },
    {
      "epoch": 0.23585381676638073,
      "grad_norm": 0.11064084959281899,
      "learning_rate": 0.00017859524959185362,
      "loss": 1.076,
      "step": 3272
    },
    {
      "epoch": 0.23592589922871765,
      "grad_norm": 0.6429080143631989,
      "learning_rate": 0.00017858081221254048,
      "loss": 1.0945,
      "step": 3273
    },
    {
      "epoch": 0.23599798169105457,
      "grad_norm": 0.24410154868614875,
      "learning_rate": 0.00017856637054987007,
      "loss": 1.0816,
      "step": 3274
    },
    {
      "epoch": 0.2360700641533915,
      "grad_norm": 0.2154320809332796,
      "learning_rate": 0.00017855192460462956,
      "loss": 1.1154,
      "step": 3275
    },
    {
      "epoch": 0.23614214661572838,
      "grad_norm": 0.29510378154277755,
      "learning_rate": 0.00017853747437760647,
      "loss": 1.1047,
      "step": 3276
    },
    {
      "epoch": 0.2362142290780653,
      "grad_norm": 0.16420445834528408,
      "learning_rate": 0.00017852301986958838,
      "loss": 1.0496,
      "step": 3277
    },
    {
      "epoch": 0.23628631154040222,
      "grad_norm": 0.36693580172644596,
      "learning_rate": 0.00017850856108136322,
      "loss": 1.0936,
      "step": 3278
    },
    {
      "epoch": 0.23635839400273914,
      "grad_norm": 0.1874755526183135,
      "learning_rate": 0.00017849409801371912,
      "loss": 1.1018,
      "step": 3279
    },
    {
      "epoch": 0.23643047646507603,
      "grad_norm": 0.17236793302504344,
      "learning_rate": 0.00017847963066744445,
      "loss": 1.0804,
      "step": 3280
    },
    {
      "epoch": 0.23650255892741295,
      "grad_norm": 0.17239513854446714,
      "learning_rate": 0.00017846515904332782,
      "loss": 1.0085,
      "step": 3281
    },
    {
      "epoch": 0.23657464138974987,
      "grad_norm": 0.17177680750677238,
      "learning_rate": 0.00017845068314215805,
      "loss": 1.0495,
      "step": 3282
    },
    {
      "epoch": 0.2366467238520868,
      "grad_norm": 2.6119647966440187,
      "learning_rate": 0.00017843620296472418,
      "loss": 1.1322,
      "step": 3283
    },
    {
      "epoch": 0.2367188063144237,
      "grad_norm": 0.11946271273420027,
      "learning_rate": 0.00017842171851181556,
      "loss": 1.2001,
      "step": 3284
    },
    {
      "epoch": 0.2367908887767606,
      "grad_norm": 0.1996555346652108,
      "learning_rate": 0.00017840722978422174,
      "loss": 1.0895,
      "step": 3285
    },
    {
      "epoch": 0.23686297123909753,
      "grad_norm": 0.2737915818172392,
      "learning_rate": 0.00017839273678273238,
      "loss": 1.1544,
      "step": 3286
    },
    {
      "epoch": 0.23693505370143444,
      "grad_norm": 0.149633184027496,
      "learning_rate": 0.00017837823950813756,
      "loss": 1.071,
      "step": 3287
    },
    {
      "epoch": 0.23700713616377136,
      "grad_norm": 0.23715781246757706,
      "learning_rate": 0.0001783637379612275,
      "loss": 1.1255,
      "step": 3288
    },
    {
      "epoch": 0.23707921862610826,
      "grad_norm": 0.22310106039758335,
      "learning_rate": 0.00017834923214279268,
      "loss": 1.1024,
      "step": 3289
    },
    {
      "epoch": 0.23715130108844518,
      "grad_norm": 1.327020762820109,
      "learning_rate": 0.00017833472205362378,
      "loss": 1.1144,
      "step": 3290
    },
    {
      "epoch": 0.2372233835507821,
      "grad_norm": 0.3099139386263716,
      "learning_rate": 0.0001783202076945117,
      "loss": 1.1428,
      "step": 3291
    },
    {
      "epoch": 0.23729546601311902,
      "grad_norm": 0.25782269696933763,
      "learning_rate": 0.00017830568906624763,
      "loss": 1.0923,
      "step": 3292
    },
    {
      "epoch": 0.2373675484754559,
      "grad_norm": 0.13717065280452692,
      "learning_rate": 0.00017829116616962297,
      "loss": 1.0717,
      "step": 3293
    },
    {
      "epoch": 0.23743963093779283,
      "grad_norm": 0.11869051975799146,
      "learning_rate": 0.00017827663900542935,
      "loss": 1.1345,
      "step": 3294
    },
    {
      "epoch": 0.23751171340012975,
      "grad_norm": 0.14482011360778338,
      "learning_rate": 0.0001782621075744586,
      "loss": 1.0847,
      "step": 3295
    },
    {
      "epoch": 0.23758379586246667,
      "grad_norm": 0.1372333224994052,
      "learning_rate": 0.00017824757187750285,
      "loss": 1.0602,
      "step": 3296
    },
    {
      "epoch": 0.2376558783248036,
      "grad_norm": 0.22640657911788595,
      "learning_rate": 0.00017823303191535442,
      "loss": 1.1655,
      "step": 3297
    },
    {
      "epoch": 0.23772796078714048,
      "grad_norm": 0.3693835533030394,
      "learning_rate": 0.00017821848768880582,
      "loss": 1.1584,
      "step": 3298
    },
    {
      "epoch": 0.2378000432494774,
      "grad_norm": 0.14226492842555957,
      "learning_rate": 0.0001782039391986499,
      "loss": 1.0891,
      "step": 3299
    },
    {
      "epoch": 0.23787212571181432,
      "grad_norm": 2.1835687431961937,
      "learning_rate": 0.00017818938644567967,
      "loss": 1.113,
      "step": 3300
    },
    {
      "epoch": 0.23794420817415124,
      "grad_norm": 0.18751744024873274,
      "learning_rate": 0.00017817482943068838,
      "loss": 1.0726,
      "step": 3301
    },
    {
      "epoch": 0.23801629063648813,
      "grad_norm": 0.16682874795907365,
      "learning_rate": 0.00017816026815446948,
      "loss": 1.073,
      "step": 3302
    },
    {
      "epoch": 0.23808837309882505,
      "grad_norm": 0.1141427683089566,
      "learning_rate": 0.0001781457026178168,
      "loss": 1.0289,
      "step": 3303
    },
    {
      "epoch": 0.23816045556116197,
      "grad_norm": 0.1676604497101875,
      "learning_rate": 0.00017813113282152414,
      "loss": 1.2028,
      "step": 3304
    },
    {
      "epoch": 0.2382325380234989,
      "grad_norm": 0.3063721954360475,
      "learning_rate": 0.00017811655876638578,
      "loss": 1.0113,
      "step": 3305
    },
    {
      "epoch": 0.23830462048583578,
      "grad_norm": 0.25797755561977,
      "learning_rate": 0.00017810198045319612,
      "loss": 1.1665,
      "step": 3306
    },
    {
      "epoch": 0.2383767029481727,
      "grad_norm": 0.12740968156103735,
      "learning_rate": 0.00017808739788274978,
      "loss": 1.0333,
      "step": 3307
    },
    {
      "epoch": 0.23844878541050962,
      "grad_norm": 0.41032704133049147,
      "learning_rate": 0.0001780728110558417,
      "loss": 1.0812,
      "step": 3308
    },
    {
      "epoch": 0.23852086787284654,
      "grad_norm": 1.2579397181596486,
      "learning_rate": 0.00017805821997326696,
      "loss": 1.1099,
      "step": 3309
    },
    {
      "epoch": 0.23859295033518346,
      "grad_norm": 0.13409611096008475,
      "learning_rate": 0.0001780436246358209,
      "loss": 1.0631,
      "step": 3310
    },
    {
      "epoch": 0.23866503279752035,
      "grad_norm": 0.1749421761080224,
      "learning_rate": 0.00017802902504429909,
      "loss": 1.0561,
      "step": 3311
    },
    {
      "epoch": 0.23873711525985727,
      "grad_norm": 0.2890339478157863,
      "learning_rate": 0.00017801442119949737,
      "loss": 1.106,
      "step": 3312
    },
    {
      "epoch": 0.2388091977221942,
      "grad_norm": 0.14620314489635625,
      "learning_rate": 0.00017799981310221173,
      "loss": 1.0575,
      "step": 3313
    },
    {
      "epoch": 0.23888128018453111,
      "grad_norm": 0.18983442119207178,
      "learning_rate": 0.00017798520075323847,
      "loss": 1.1176,
      "step": 3314
    },
    {
      "epoch": 0.238953362646868,
      "grad_norm": 0.3691140465826691,
      "learning_rate": 0.0001779705841533741,
      "loss": 1.0744,
      "step": 3315
    },
    {
      "epoch": 0.23902544510920493,
      "grad_norm": 0.18887206542968923,
      "learning_rate": 0.00017795596330341532,
      "loss": 1.1155,
      "step": 3316
    },
    {
      "epoch": 0.23909752757154185,
      "grad_norm": 0.12253974852346124,
      "learning_rate": 0.00017794133820415916,
      "loss": 1.0851,
      "step": 3317
    },
    {
      "epoch": 0.23916961003387877,
      "grad_norm": 0.15508880101803782,
      "learning_rate": 0.0001779267088564028,
      "loss": 1.1325,
      "step": 3318
    },
    {
      "epoch": 0.23924169249621566,
      "grad_norm": 0.36973798278867,
      "learning_rate": 0.00017791207526094362,
      "loss": 1.1036,
      "step": 3319
    },
    {
      "epoch": 0.23931377495855258,
      "grad_norm": 0.15189796873454317,
      "learning_rate": 0.00017789743741857933,
      "loss": 1.1281,
      "step": 3320
    },
    {
      "epoch": 0.2393858574208895,
      "grad_norm": 0.29825428461027703,
      "learning_rate": 0.00017788279533010785,
      "loss": 1.1621,
      "step": 3321
    },
    {
      "epoch": 0.23945793988322642,
      "grad_norm": 0.19643740315075775,
      "learning_rate": 0.0001778681489963272,
      "loss": 1.1001,
      "step": 3322
    },
    {
      "epoch": 0.23953002234556334,
      "grad_norm": 0.275523925439647,
      "learning_rate": 0.00017785349841803583,
      "loss": 1.123,
      "step": 3323
    },
    {
      "epoch": 0.23960210480790023,
      "grad_norm": 0.2857861104119372,
      "learning_rate": 0.0001778388435960323,
      "loss": 1.1892,
      "step": 3324
    },
    {
      "epoch": 0.23967418727023715,
      "grad_norm": 0.11606950255217681,
      "learning_rate": 0.0001778241845311154,
      "loss": 1.1129,
      "step": 3325
    },
    {
      "epoch": 0.23974626973257407,
      "grad_norm": 0.14838431204544708,
      "learning_rate": 0.00017780952122408423,
      "loss": 1.1061,
      "step": 3326
    },
    {
      "epoch": 0.239818352194911,
      "grad_norm": 1.210771559641689,
      "learning_rate": 0.00017779485367573806,
      "loss": 1.1118,
      "step": 3327
    },
    {
      "epoch": 0.23989043465724788,
      "grad_norm": 0.23703265235304205,
      "learning_rate": 0.0001777801818868764,
      "loss": 1.0563,
      "step": 3328
    },
    {
      "epoch": 0.2399625171195848,
      "grad_norm": 0.16852776995893406,
      "learning_rate": 0.00017776550585829896,
      "loss": 1.1162,
      "step": 3329
    },
    {
      "epoch": 0.24003459958192172,
      "grad_norm": 0.14400997730092524,
      "learning_rate": 0.00017775082559080575,
      "loss": 1.1145,
      "step": 3330
    },
    {
      "epoch": 0.24010668204425864,
      "grad_norm": 0.14335301867411152,
      "learning_rate": 0.00017773614108519697,
      "loss": 1.0088,
      "step": 3331
    },
    {
      "epoch": 0.24017876450659553,
      "grad_norm": 0.18597156237959075,
      "learning_rate": 0.00017772145234227306,
      "loss": 1.0704,
      "step": 3332
    },
    {
      "epoch": 0.24025084696893245,
      "grad_norm": 0.3191376119663256,
      "learning_rate": 0.00017770675936283465,
      "loss": 1.0696,
      "step": 3333
    },
    {
      "epoch": 0.24032292943126937,
      "grad_norm": 0.12216690546963967,
      "learning_rate": 0.0001776920621476827,
      "loss": 1.0599,
      "step": 3334
    },
    {
      "epoch": 0.2403950118936063,
      "grad_norm": 0.23552018860101456,
      "learning_rate": 0.0001776773606976183,
      "loss": 1.1246,
      "step": 3335
    },
    {
      "epoch": 0.2404670943559432,
      "grad_norm": 0.15581881815298856,
      "learning_rate": 0.00017766265501344283,
      "loss": 1.161,
      "step": 3336
    },
    {
      "epoch": 0.2405391768182801,
      "grad_norm": 0.3905817978038121,
      "learning_rate": 0.00017764794509595786,
      "loss": 1.0399,
      "step": 3337
    },
    {
      "epoch": 0.24061125928061702,
      "grad_norm": 0.12743929279395708,
      "learning_rate": 0.00017763323094596522,
      "loss": 0.9954,
      "step": 3338
    },
    {
      "epoch": 0.24068334174295394,
      "grad_norm": 0.20526785680624707,
      "learning_rate": 0.00017761851256426698,
      "loss": 1.0975,
      "step": 3339
    },
    {
      "epoch": 0.24075542420529086,
      "grad_norm": 0.37777815679255866,
      "learning_rate": 0.00017760378995166541,
      "loss": 1.0943,
      "step": 3340
    },
    {
      "epoch": 0.24082750666762776,
      "grad_norm": 0.26018597552179373,
      "learning_rate": 0.000177589063108963,
      "loss": 1.0733,
      "step": 3341
    },
    {
      "epoch": 0.24089958912996468,
      "grad_norm": 0.28615848015590284,
      "learning_rate": 0.00017757433203696255,
      "loss": 1.0926,
      "step": 3342
    },
    {
      "epoch": 0.2409716715923016,
      "grad_norm": 0.1265702145318758,
      "learning_rate": 0.00017755959673646697,
      "loss": 1.0576,
      "step": 3343
    },
    {
      "epoch": 0.24104375405463851,
      "grad_norm": 0.12263843867487179,
      "learning_rate": 0.00017754485720827953,
      "loss": 1.1492,
      "step": 3344
    },
    {
      "epoch": 0.2411158365169754,
      "grad_norm": 0.2881694051847863,
      "learning_rate": 0.00017753011345320366,
      "loss": 1.1091,
      "step": 3345
    },
    {
      "epoch": 0.24118791897931233,
      "grad_norm": 0.2734937536279754,
      "learning_rate": 0.00017751536547204295,
      "loss": 1.0906,
      "step": 3346
    },
    {
      "epoch": 0.24126000144164925,
      "grad_norm": 0.17762376738522595,
      "learning_rate": 0.00017750061326560136,
      "loss": 1.1238,
      "step": 3347
    },
    {
      "epoch": 0.24133208390398617,
      "grad_norm": 0.814229773546474,
      "learning_rate": 0.00017748585683468302,
      "loss": 1.125,
      "step": 3348
    },
    {
      "epoch": 0.2414041663663231,
      "grad_norm": 0.20293977572428162,
      "learning_rate": 0.00017747109618009227,
      "loss": 1.0721,
      "step": 3349
    },
    {
      "epoch": 0.24147624882865998,
      "grad_norm": 1.7142616060013116,
      "learning_rate": 0.0001774563313026337,
      "loss": 1.0316,
      "step": 3350
    },
    {
      "epoch": 0.2415483312909969,
      "grad_norm": 0.2766167797653695,
      "learning_rate": 0.00017744156220311216,
      "loss": 1.193,
      "step": 3351
    },
    {
      "epoch": 0.24162041375333382,
      "grad_norm": 0.22264968635498683,
      "learning_rate": 0.00017742678888233262,
      "loss": 1.0595,
      "step": 3352
    },
    {
      "epoch": 0.24169249621567074,
      "grad_norm": 0.13308614953943673,
      "learning_rate": 0.00017741201134110042,
      "loss": 1.0599,
      "step": 3353
    },
    {
      "epoch": 0.24176457867800763,
      "grad_norm": 0.2555094177419172,
      "learning_rate": 0.00017739722958022108,
      "loss": 1.1046,
      "step": 3354
    },
    {
      "epoch": 0.24183666114034455,
      "grad_norm": 0.10623222059768703,
      "learning_rate": 0.0001773824436005003,
      "loss": 1.0813,
      "step": 3355
    },
    {
      "epoch": 0.24190874360268147,
      "grad_norm": 0.5846329016833377,
      "learning_rate": 0.00017736765340274407,
      "loss": 1.1446,
      "step": 3356
    },
    {
      "epoch": 0.2419808260650184,
      "grad_norm": 0.12562294450350292,
      "learning_rate": 0.00017735285898775853,
      "loss": 1.0075,
      "step": 3357
    },
    {
      "epoch": 0.24205290852735528,
      "grad_norm": 0.1157242844742407,
      "learning_rate": 0.00017733806035635018,
      "loss": 1.0485,
      "step": 3358
    },
    {
      "epoch": 0.2421249909896922,
      "grad_norm": 0.6437082903304178,
      "learning_rate": 0.00017732325750932567,
      "loss": 1.0968,
      "step": 3359
    },
    {
      "epoch": 0.24219707345202912,
      "grad_norm": 0.20924667732659388,
      "learning_rate": 0.00017730845044749188,
      "loss": 1.0355,
      "step": 3360
    },
    {
      "epoch": 0.24226915591436604,
      "grad_norm": 0.18580510455769889,
      "learning_rate": 0.0001772936391716559,
      "loss": 1.0909,
      "step": 3361
    },
    {
      "epoch": 0.24234123837670296,
      "grad_norm": 0.11115644687239001,
      "learning_rate": 0.0001772788236826251,
      "loss": 1.0445,
      "step": 3362
    },
    {
      "epoch": 0.24241332083903985,
      "grad_norm": 0.10874914608417714,
      "learning_rate": 0.00017726400398120705,
      "loss": 1.1121,
      "step": 3363
    },
    {
      "epoch": 0.24248540330137677,
      "grad_norm": 0.18731283862192105,
      "learning_rate": 0.0001772491800682096,
      "loss": 1.0839,
      "step": 3364
    },
    {
      "epoch": 0.2425574857637137,
      "grad_norm": 0.9881648064566324,
      "learning_rate": 0.0001772343519444407,
      "loss": 1.0851,
      "step": 3365
    },
    {
      "epoch": 0.2426295682260506,
      "grad_norm": 0.23431657264612585,
      "learning_rate": 0.00017721951961070865,
      "loss": 1.1007,
      "step": 3366
    },
    {
      "epoch": 0.2427016506883875,
      "grad_norm": 0.11848595688638973,
      "learning_rate": 0.00017720468306782196,
      "loss": 1.1205,
      "step": 3367
    },
    {
      "epoch": 0.24277373315072442,
      "grad_norm": 0.15548434705224773,
      "learning_rate": 0.00017718984231658938,
      "loss": 1.0322,
      "step": 3368
    },
    {
      "epoch": 0.24284581561306134,
      "grad_norm": 0.2728993552100176,
      "learning_rate": 0.00017717499735781983,
      "loss": 1.0794,
      "step": 3369
    },
    {
      "epoch": 0.24291789807539826,
      "grad_norm": 0.132732632313711,
      "learning_rate": 0.0001771601481923225,
      "loss": 1.074,
      "step": 3370
    },
    {
      "epoch": 0.24298998053773516,
      "grad_norm": 0.1297292223839261,
      "learning_rate": 0.00017714529482090678,
      "loss": 1.1072,
      "step": 3371
    },
    {
      "epoch": 0.24306206300007208,
      "grad_norm": 0.2220205096916586,
      "learning_rate": 0.0001771304372443823,
      "loss": 1.0285,
      "step": 3372
    },
    {
      "epoch": 0.243134145462409,
      "grad_norm": 0.1697645121118699,
      "learning_rate": 0.00017711557546355904,
      "loss": 1.0387,
      "step": 3373
    },
    {
      "epoch": 0.24320622792474592,
      "grad_norm": 0.12766622204933523,
      "learning_rate": 0.00017710070947924697,
      "loss": 1.0755,
      "step": 3374
    },
    {
      "epoch": 0.24327831038708284,
      "grad_norm": 0.2569579015634898,
      "learning_rate": 0.0001770858392922565,
      "loss": 1.0708,
      "step": 3375
    },
    {
      "epoch": 0.24335039284941973,
      "grad_norm": 0.26071110142201304,
      "learning_rate": 0.0001770709649033981,
      "loss": 1.1189,
      "step": 3376
    },
    {
      "epoch": 0.24342247531175665,
      "grad_norm": 0.13238596792022414,
      "learning_rate": 0.00017705608631348268,
      "loss": 1.1376,
      "step": 3377
    },
    {
      "epoch": 0.24349455777409357,
      "grad_norm": 0.142045448115027,
      "learning_rate": 0.00017704120352332118,
      "loss": 1.0389,
      "step": 3378
    },
    {
      "epoch": 0.2435666402364305,
      "grad_norm": 0.10885109247792808,
      "learning_rate": 0.00017702631653372485,
      "loss": 1.0275,
      "step": 3379
    },
    {
      "epoch": 0.24363872269876738,
      "grad_norm": 0.22735117724949738,
      "learning_rate": 0.00017701142534550513,
      "loss": 1.1171,
      "step": 3380
    },
    {
      "epoch": 0.2437108051611043,
      "grad_norm": 0.20157321142341353,
      "learning_rate": 0.00017699652995947383,
      "loss": 1.0759,
      "step": 3381
    },
    {
      "epoch": 0.24378288762344122,
      "grad_norm": 0.144010763219303,
      "learning_rate": 0.00017698163037644277,
      "loss": 1.0436,
      "step": 3382
    },
    {
      "epoch": 0.24385497008577814,
      "grad_norm": 0.14415693986230002,
      "learning_rate": 0.00017696672659722417,
      "loss": 1.0818,
      "step": 3383
    },
    {
      "epoch": 0.24392705254811503,
      "grad_norm": 0.17480786707295867,
      "learning_rate": 0.00017695181862263044,
      "loss": 1.0886,
      "step": 3384
    },
    {
      "epoch": 0.24399913501045195,
      "grad_norm": 0.1882041876747594,
      "learning_rate": 0.00017693690645347414,
      "loss": 1.0505,
      "step": 3385
    },
    {
      "epoch": 0.24407121747278887,
      "grad_norm": 0.11356711485516371,
      "learning_rate": 0.00017692199009056815,
      "loss": 1.0951,
      "step": 3386
    },
    {
      "epoch": 0.2441432999351258,
      "grad_norm": 0.12211255677024585,
      "learning_rate": 0.00017690706953472552,
      "loss": 1.0159,
      "step": 3387
    },
    {
      "epoch": 0.2442153823974627,
      "grad_norm": 0.1947490049486219,
      "learning_rate": 0.00017689214478675954,
      "loss": 1.1415,
      "step": 3388
    },
    {
      "epoch": 0.2442874648597996,
      "grad_norm": 0.11988950780640441,
      "learning_rate": 0.00017687721584748383,
      "loss": 1.1541,
      "step": 3389
    },
    {
      "epoch": 0.24435954732213652,
      "grad_norm": 1.929350327023143,
      "learning_rate": 0.00017686228271771205,
      "loss": 1.0894,
      "step": 3390
    },
    {
      "epoch": 0.24443162978447344,
      "grad_norm": 0.15007309376978917,
      "learning_rate": 0.00017684734539825825,
      "loss": 1.1523,
      "step": 3391
    },
    {
      "epoch": 0.24450371224681036,
      "grad_norm": 0.20050124860934754,
      "learning_rate": 0.00017683240388993664,
      "loss": 1.0426,
      "step": 3392
    },
    {
      "epoch": 0.24457579470914725,
      "grad_norm": 0.36055673004879335,
      "learning_rate": 0.00017681745819356164,
      "loss": 1.0979,
      "step": 3393
    },
    {
      "epoch": 0.24464787717148417,
      "grad_norm": 0.11974798106106423,
      "learning_rate": 0.00017680250830994794,
      "loss": 1.0428,
      "step": 3394
    },
    {
      "epoch": 0.2447199596338211,
      "grad_norm": 0.17185168290170186,
      "learning_rate": 0.0001767875542399105,
      "loss": 1.1244,
      "step": 3395
    },
    {
      "epoch": 0.244792042096158,
      "grad_norm": 0.23323329392666947,
      "learning_rate": 0.00017677259598426433,
      "loss": 1.0962,
      "step": 3396
    },
    {
      "epoch": 0.2448641245584949,
      "grad_norm": 0.13968005169495662,
      "learning_rate": 0.00017675763354382484,
      "loss": 1.0332,
      "step": 3397
    },
    {
      "epoch": 0.24493620702083183,
      "grad_norm": 0.19089493388667653,
      "learning_rate": 0.00017674266691940766,
      "loss": 1.1025,
      "step": 3398
    },
    {
      "epoch": 0.24500828948316875,
      "grad_norm": 0.12833281759230938,
      "learning_rate": 0.00017672769611182858,
      "loss": 1.1153,
      "step": 3399
    },
    {
      "epoch": 0.24508037194550567,
      "grad_norm": 0.12598895553910247,
      "learning_rate": 0.0001767127211219036,
      "loss": 1.0507,
      "step": 3400
    },
    {
      "epoch": 0.24515245440784258,
      "grad_norm": 0.3397511043701996,
      "learning_rate": 0.00017669774195044907,
      "loss": 1.0439,
      "step": 3401
    },
    {
      "epoch": 0.24522453687017948,
      "grad_norm": 0.17776969680115295,
      "learning_rate": 0.00017668275859828141,
      "loss": 1.0882,
      "step": 3402
    },
    {
      "epoch": 0.2452966193325164,
      "grad_norm": 0.1275371478016765,
      "learning_rate": 0.00017666777106621742,
      "loss": 1.0651,
      "step": 3403
    },
    {
      "epoch": 0.24536870179485332,
      "grad_norm": 0.11640035732555347,
      "learning_rate": 0.00017665277935507398,
      "loss": 1.0225,
      "step": 3404
    },
    {
      "epoch": 0.24544078425719024,
      "grad_norm": 0.13101935638375997,
      "learning_rate": 0.00017663778346566831,
      "loss": 1.0484,
      "step": 3405
    },
    {
      "epoch": 0.24551286671952713,
      "grad_norm": 0.1851981317524214,
      "learning_rate": 0.00017662278339881785,
      "loss": 1.0872,
      "step": 3406
    },
    {
      "epoch": 0.24558494918186405,
      "grad_norm": 0.12801491813874252,
      "learning_rate": 0.00017660777915534022,
      "loss": 1.1425,
      "step": 3407
    },
    {
      "epoch": 0.24565703164420097,
      "grad_norm": 0.15680463229773625,
      "learning_rate": 0.00017659277073605323,
      "loss": 1.0878,
      "step": 3408
    },
    {
      "epoch": 0.2457291141065379,
      "grad_norm": 0.2603907326637543,
      "learning_rate": 0.00017657775814177505,
      "loss": 1.1014,
      "step": 3409
    },
    {
      "epoch": 0.24580119656887478,
      "grad_norm": 0.09795479375418191,
      "learning_rate": 0.0001765627413733239,
      "loss": 1.0338,
      "step": 3410
    },
    {
      "epoch": 0.2458732790312117,
      "grad_norm": 0.12160414060897261,
      "learning_rate": 0.00017654772043151847,
      "loss": 1.0787,
      "step": 3411
    },
    {
      "epoch": 0.24594536149354862,
      "grad_norm": 0.1266873300655095,
      "learning_rate": 0.00017653269531717745,
      "loss": 1.1399,
      "step": 3412
    },
    {
      "epoch": 0.24601744395588554,
      "grad_norm": 0.18623111031520687,
      "learning_rate": 0.00017651766603111983,
      "loss": 1.0236,
      "step": 3413
    },
    {
      "epoch": 0.24608952641822246,
      "grad_norm": 0.17261718420990935,
      "learning_rate": 0.00017650263257416485,
      "loss": 1.1178,
      "step": 3414
    },
    {
      "epoch": 0.24616160888055935,
      "grad_norm": 0.12272049705417991,
      "learning_rate": 0.00017648759494713203,
      "loss": 1.1199,
      "step": 3415
    },
    {
      "epoch": 0.24623369134289627,
      "grad_norm": 0.16254397630604625,
      "learning_rate": 0.000176472553150841,
      "loss": 1.0627,
      "step": 3416
    },
    {
      "epoch": 0.2463057738052332,
      "grad_norm": 0.12522168832911942,
      "learning_rate": 0.00017645750718611164,
      "loss": 1.1465,
      "step": 3417
    },
    {
      "epoch": 0.2463778562675701,
      "grad_norm": 0.20409634100973664,
      "learning_rate": 0.00017644245705376416,
      "loss": 1.1549,
      "step": 3418
    },
    {
      "epoch": 0.246449938729907,
      "grad_norm": 0.1485069370533696,
      "learning_rate": 0.0001764274027546189,
      "loss": 1.1082,
      "step": 3419
    },
    {
      "epoch": 0.24652202119224392,
      "grad_norm": 0.10798149502424602,
      "learning_rate": 0.00017641234428949644,
      "loss": 1.0159,
      "step": 3420
    },
    {
      "epoch": 0.24659410365458084,
      "grad_norm": 0.457406090843671,
      "learning_rate": 0.00017639728165921764,
      "loss": 1.1494,
      "step": 3421
    },
    {
      "epoch": 0.24666618611691776,
      "grad_norm": 0.13787033639627455,
      "learning_rate": 0.00017638221486460348,
      "loss": 1.0438,
      "step": 3422
    },
    {
      "epoch": 0.24673826857925465,
      "grad_norm": 0.21433863233638192,
      "learning_rate": 0.0001763671439064753,
      "loss": 1.0238,
      "step": 3423
    },
    {
      "epoch": 0.24681035104159157,
      "grad_norm": 0.15600571887978937,
      "learning_rate": 0.00017635206878565456,
      "loss": 1.1306,
      "step": 3424
    },
    {
      "epoch": 0.2468824335039285,
      "grad_norm": 0.13941049450386322,
      "learning_rate": 0.000176336989502963,
      "loss": 1.0265,
      "step": 3425
    },
    {
      "epoch": 0.24695451596626541,
      "grad_norm": 0.48205066123254103,
      "learning_rate": 0.0001763219060592226,
      "loss": 1.1623,
      "step": 3426
    },
    {
      "epoch": 0.24702659842860233,
      "grad_norm": 0.21625233069194166,
      "learning_rate": 0.00017630681845525552,
      "loss": 1.1423,
      "step": 3427
    },
    {
      "epoch": 0.24709868089093923,
      "grad_norm": 0.17092594959987864,
      "learning_rate": 0.00017629172669188418,
      "loss": 1.131,
      "step": 3428
    },
    {
      "epoch": 0.24717076335327615,
      "grad_norm": 0.1301511929158189,
      "learning_rate": 0.0001762766307699312,
      "loss": 1.0572,
      "step": 3429
    },
    {
      "epoch": 0.24724284581561307,
      "grad_norm": 0.23703098952315096,
      "learning_rate": 0.00017626153069021946,
      "loss": 1.1022,
      "step": 3430
    },
    {
      "epoch": 0.24731492827794999,
      "grad_norm": 0.13582086379182534,
      "learning_rate": 0.00017624642645357205,
      "loss": 1.0309,
      "step": 3431
    },
    {
      "epoch": 0.24738701074028688,
      "grad_norm": 0.19534204424375443,
      "learning_rate": 0.00017623131806081225,
      "loss": 1.0227,
      "step": 3432
    },
    {
      "epoch": 0.2474590932026238,
      "grad_norm": 0.16725224826218923,
      "learning_rate": 0.00017621620551276366,
      "loss": 1.1503,
      "step": 3433
    },
    {
      "epoch": 0.24753117566496072,
      "grad_norm": 0.14218186142128314,
      "learning_rate": 0.00017620108881025002,
      "loss": 1.1226,
      "step": 3434
    },
    {
      "epoch": 0.24760325812729764,
      "grad_norm": 0.2360146121911134,
      "learning_rate": 0.0001761859679540953,
      "loss": 1.0215,
      "step": 3435
    },
    {
      "epoch": 0.24767534058963453,
      "grad_norm": 0.3750174765003825,
      "learning_rate": 0.00017617084294512377,
      "loss": 1.1155,
      "step": 3436
    },
    {
      "epoch": 0.24774742305197145,
      "grad_norm": 0.1475712774642911,
      "learning_rate": 0.00017615571378415984,
      "loss": 1.0947,
      "step": 3437
    },
    {
      "epoch": 0.24781950551430837,
      "grad_norm": 0.1411704799780267,
      "learning_rate": 0.0001761405804720282,
      "loss": 1.0731,
      "step": 3438
    },
    {
      "epoch": 0.2478915879766453,
      "grad_norm": 0.15099483944403785,
      "learning_rate": 0.00017612544300955375,
      "loss": 1.04,
      "step": 3439
    },
    {
      "epoch": 0.2479636704389822,
      "grad_norm": 0.3849924101642859,
      "learning_rate": 0.00017611030139756163,
      "loss": 1.1376,
      "step": 3440
    },
    {
      "epoch": 0.2480357529013191,
      "grad_norm": 0.12040469509161446,
      "learning_rate": 0.0001760951556368772,
      "loss": 1.096,
      "step": 3441
    },
    {
      "epoch": 0.24810783536365602,
      "grad_norm": 0.1560228457297703,
      "learning_rate": 0.00017608000572832597,
      "loss": 1.0913,
      "step": 3442
    },
    {
      "epoch": 0.24817991782599294,
      "grad_norm": 0.3226127485759522,
      "learning_rate": 0.00017606485167273383,
      "loss": 0.9681,
      "step": 3443
    },
    {
      "epoch": 0.24825200028832986,
      "grad_norm": 0.44691526066002657,
      "learning_rate": 0.00017604969347092677,
      "loss": 1.0502,
      "step": 3444
    },
    {
      "epoch": 0.24832408275066675,
      "grad_norm": 0.11694042844758823,
      "learning_rate": 0.00017603453112373108,
      "loss": 1.0961,
      "step": 3445
    },
    {
      "epoch": 0.24839616521300367,
      "grad_norm": 0.201125841648338,
      "learning_rate": 0.00017601936463197318,
      "loss": 1.1279,
      "step": 3446
    },
    {
      "epoch": 0.2484682476753406,
      "grad_norm": 0.14118833632032396,
      "learning_rate": 0.00017600419399647983,
      "loss": 1.0186,
      "step": 3447
    },
    {
      "epoch": 0.2485403301376775,
      "grad_norm": 0.19344295668683367,
      "learning_rate": 0.000175989019218078,
      "loss": 1.0986,
      "step": 3448
    },
    {
      "epoch": 0.2486124126000144,
      "grad_norm": 0.22676840824961433,
      "learning_rate": 0.00017597384029759475,
      "loss": 1.0388,
      "step": 3449
    },
    {
      "epoch": 0.24868449506235132,
      "grad_norm": 0.2618165633818039,
      "learning_rate": 0.00017595865723585755,
      "loss": 1.0297,
      "step": 3450
    },
    {
      "epoch": 0.24875657752468824,
      "grad_norm": 0.26933167263680213,
      "learning_rate": 0.00017594347003369395,
      "loss": 1.0563,
      "step": 3451
    },
    {
      "epoch": 0.24882865998702516,
      "grad_norm": 0.1270698845265757,
      "learning_rate": 0.00017592827869193184,
      "loss": 1.0081,
      "step": 3452
    },
    {
      "epoch": 0.24890074244936208,
      "grad_norm": 0.2717572035296911,
      "learning_rate": 0.00017591308321139929,
      "loss": 1.0682,
      "step": 3453
    },
    {
      "epoch": 0.24897282491169898,
      "grad_norm": 0.15740808106414933,
      "learning_rate": 0.00017589788359292455,
      "loss": 1.0935,
      "step": 3454
    },
    {
      "epoch": 0.2490449073740359,
      "grad_norm": 0.2837893195815509,
      "learning_rate": 0.00017588267983733617,
      "loss": 1.0974,
      "step": 3455
    },
    {
      "epoch": 0.24911698983637282,
      "grad_norm": 0.16343445288771186,
      "learning_rate": 0.00017586747194546284,
      "loss": 1.1098,
      "step": 3456
    },
    {
      "epoch": 0.24918907229870974,
      "grad_norm": 0.1923791020780723,
      "learning_rate": 0.0001758522599181336,
      "loss": 1.0562,
      "step": 3457
    },
    {
      "epoch": 0.24926115476104663,
      "grad_norm": 0.17165237391248606,
      "learning_rate": 0.00017583704375617756,
      "loss": 1.1148,
      "step": 3458
    },
    {
      "epoch": 0.24933323722338355,
      "grad_norm": 0.23727385150766303,
      "learning_rate": 0.00017582182346042422,
      "loss": 1.0869,
      "step": 3459
    },
    {
      "epoch": 0.24940531968572047,
      "grad_norm": 0.21535007435695966,
      "learning_rate": 0.00017580659903170312,
      "loss": 1.095,
      "step": 3460
    },
    {
      "epoch": 0.2494774021480574,
      "grad_norm": 0.18388852370112185,
      "learning_rate": 0.00017579137047084425,
      "loss": 1.1018,
      "step": 3461
    },
    {
      "epoch": 0.24954948461039428,
      "grad_norm": 0.11803365613794511,
      "learning_rate": 0.00017577613777867762,
      "loss": 1.0009,
      "step": 3462
    },
    {
      "epoch": 0.2496215670727312,
      "grad_norm": 0.17283184728273737,
      "learning_rate": 0.00017576090095603356,
      "loss": 1.0644,
      "step": 3463
    },
    {
      "epoch": 0.24969364953506812,
      "grad_norm": 0.14263137166039377,
      "learning_rate": 0.00017574566000374262,
      "loss": 1.1085,
      "step": 3464
    },
    {
      "epoch": 0.24976573199740504,
      "grad_norm": 0.15965960732221293,
      "learning_rate": 0.00017573041492263558,
      "loss": 1.0136,
      "step": 3465
    },
    {
      "epoch": 0.24983781445974196,
      "grad_norm": 0.12547492177948405,
      "learning_rate": 0.0001757151657135434,
      "loss": 1.0841,
      "step": 3466
    },
    {
      "epoch": 0.24990989692207885,
      "grad_norm": 0.1254205847247981,
      "learning_rate": 0.00017569991237729733,
      "loss": 1.0705,
      "step": 3467
    },
    {
      "epoch": 0.24998197938441577,
      "grad_norm": 0.1738242642097683,
      "learning_rate": 0.00017568465491472882,
      "loss": 1.1194,
      "step": 3468
    },
    {
      "epoch": 0.25005406184675266,
      "grad_norm": 0.1535597368395432,
      "learning_rate": 0.0001756693933266695,
      "loss": 1.0558,
      "step": 3469
    },
    {
      "epoch": 0.2501261443090896,
      "grad_norm": 0.35060473553083443,
      "learning_rate": 0.00017565412761395133,
      "loss": 1.0628,
      "step": 3470
    },
    {
      "epoch": 0.2501982267714265,
      "grad_norm": 0.12265793685634328,
      "learning_rate": 0.00017563885777740636,
      "loss": 1.0803,
      "step": 3471
    },
    {
      "epoch": 0.2502703092337634,
      "grad_norm": 0.31662994650953724,
      "learning_rate": 0.00017562358381786693,
      "loss": 1.0439,
      "step": 3472
    },
    {
      "epoch": 0.25034239169610034,
      "grad_norm": 0.11962093519896702,
      "learning_rate": 0.00017560830573616566,
      "loss": 1.0179,
      "step": 3473
    },
    {
      "epoch": 0.25041447415843726,
      "grad_norm": 0.1359833130795473,
      "learning_rate": 0.0001755930235331353,
      "loss": 1.1278,
      "step": 3474
    },
    {
      "epoch": 0.2504865566207742,
      "grad_norm": 0.13056884324341145,
      "learning_rate": 0.00017557773720960893,
      "loss": 1.1239,
      "step": 3475
    },
    {
      "epoch": 0.2505586390831111,
      "grad_norm": 0.1551032499332178,
      "learning_rate": 0.0001755624467664197,
      "loss": 1.081,
      "step": 3476
    },
    {
      "epoch": 0.25063072154544797,
      "grad_norm": 0.892400624602093,
      "learning_rate": 0.00017554715220440112,
      "loss": 1.1508,
      "step": 3477
    },
    {
      "epoch": 0.2507028040077849,
      "grad_norm": 0.20791525697888558,
      "learning_rate": 0.0001755318535243869,
      "loss": 1.099,
      "step": 3478
    },
    {
      "epoch": 0.2507748864701218,
      "grad_norm": 0.17703849550764364,
      "learning_rate": 0.00017551655072721092,
      "loss": 1.0438,
      "step": 3479
    },
    {
      "epoch": 0.2508469689324587,
      "grad_norm": 0.1689525155859107,
      "learning_rate": 0.00017550124381370734,
      "loss": 0.9717,
      "step": 3480
    },
    {
      "epoch": 0.25091905139479564,
      "grad_norm": 0.13542837114618989,
      "learning_rate": 0.00017548593278471052,
      "loss": 1.0037,
      "step": 3481
    },
    {
      "epoch": 0.25099113385713256,
      "grad_norm": 0.11656646811678711,
      "learning_rate": 0.00017547061764105503,
      "loss": 1.1159,
      "step": 3482
    },
    {
      "epoch": 0.2510632163194695,
      "grad_norm": 0.3516536437174794,
      "learning_rate": 0.00017545529838357572,
      "loss": 1.0813,
      "step": 3483
    },
    {
      "epoch": 0.2511352987818064,
      "grad_norm": 0.1276584468167134,
      "learning_rate": 0.00017543997501310763,
      "loss": 1.0804,
      "step": 3484
    },
    {
      "epoch": 0.2512073812441433,
      "grad_norm": 0.11832688896673121,
      "learning_rate": 0.00017542464753048594,
      "loss": 1.0497,
      "step": 3485
    },
    {
      "epoch": 0.2512794637064802,
      "grad_norm": 0.12468483584876741,
      "learning_rate": 0.00017540931593654623,
      "loss": 1.0486,
      "step": 3486
    },
    {
      "epoch": 0.2513515461688171,
      "grad_norm": 1.4981750074909082,
      "learning_rate": 0.00017539398023212415,
      "loss": 1.0995,
      "step": 3487
    },
    {
      "epoch": 0.25142362863115403,
      "grad_norm": 0.23766636352888673,
      "learning_rate": 0.00017537864041805568,
      "loss": 0.9846,
      "step": 3488
    },
    {
      "epoch": 0.25149571109349095,
      "grad_norm": 0.2247910252542719,
      "learning_rate": 0.00017536329649517693,
      "loss": 1.0522,
      "step": 3489
    },
    {
      "epoch": 0.25156779355582787,
      "grad_norm": 0.19061159290679655,
      "learning_rate": 0.0001753479484643243,
      "loss": 1.0999,
      "step": 3490
    },
    {
      "epoch": 0.2516398760181648,
      "grad_norm": 0.20737737538016676,
      "learning_rate": 0.00017533259632633442,
      "loss": 1.1017,
      "step": 3491
    },
    {
      "epoch": 0.2517119584805017,
      "grad_norm": 0.1314706417468263,
      "learning_rate": 0.00017531724008204406,
      "loss": 1.0338,
      "step": 3492
    },
    {
      "epoch": 0.2517840409428386,
      "grad_norm": 0.13384867472378323,
      "learning_rate": 0.00017530187973229032,
      "loss": 1.0616,
      "step": 3493
    },
    {
      "epoch": 0.25185612340517555,
      "grad_norm": 0.15632042643535216,
      "learning_rate": 0.00017528651527791048,
      "loss": 1.0718,
      "step": 3494
    },
    {
      "epoch": 0.2519282058675124,
      "grad_norm": 0.11109378249921599,
      "learning_rate": 0.000175271146719742,
      "loss": 1.0624,
      "step": 3495
    },
    {
      "epoch": 0.25200028832984933,
      "grad_norm": 0.6846575387933671,
      "learning_rate": 0.00017525577405862265,
      "loss": 1.0515,
      "step": 3496
    },
    {
      "epoch": 0.25207237079218625,
      "grad_norm": 0.12978775350882718,
      "learning_rate": 0.00017524039729539036,
      "loss": 1.0466,
      "step": 3497
    },
    {
      "epoch": 0.25214445325452317,
      "grad_norm": 0.43841310780546844,
      "learning_rate": 0.00017522501643088328,
      "loss": 1.1189,
      "step": 3498
    },
    {
      "epoch": 0.2522165357168601,
      "grad_norm": 0.12183764484768549,
      "learning_rate": 0.00017520963146593985,
      "loss": 1.0502,
      "step": 3499
    },
    {
      "epoch": 0.252288618179197,
      "grad_norm": 0.14537909802242865,
      "learning_rate": 0.00017519424240139864,
      "loss": 1.0761,
      "step": 3500
    },
    {
      "epoch": 0.25236070064153393,
      "grad_norm": 0.11538978244014308,
      "learning_rate": 0.0001751788492380985,
      "loss": 1.0211,
      "step": 3501
    },
    {
      "epoch": 0.25243278310387085,
      "grad_norm": 0.12098159294575164,
      "learning_rate": 0.00017516345197687852,
      "loss": 1.0372,
      "step": 3502
    },
    {
      "epoch": 0.2525048655662077,
      "grad_norm": 0.16410331055167113,
      "learning_rate": 0.00017514805061857798,
      "loss": 1.0714,
      "step": 3503
    },
    {
      "epoch": 0.25257694802854463,
      "grad_norm": 0.21204994461785026,
      "learning_rate": 0.00017513264516403641,
      "loss": 1.0419,
      "step": 3504
    },
    {
      "epoch": 0.25264903049088155,
      "grad_norm": 0.168305695723436,
      "learning_rate": 0.00017511723561409347,
      "loss": 1.1499,
      "step": 3505
    },
    {
      "epoch": 0.2527211129532185,
      "grad_norm": 0.7168794823726422,
      "learning_rate": 0.0001751018219695892,
      "loss": 1.0384,
      "step": 3506
    },
    {
      "epoch": 0.2527931954155554,
      "grad_norm": 0.16168045513939866,
      "learning_rate": 0.00017508640423136377,
      "loss": 1.1455,
      "step": 3507
    },
    {
      "epoch": 0.2528652778778923,
      "grad_norm": 0.19821556168942678,
      "learning_rate": 0.0001750709824002575,
      "loss": 1.1418,
      "step": 3508
    },
    {
      "epoch": 0.25293736034022923,
      "grad_norm": 0.24444566770662565,
      "learning_rate": 0.00017505555647711116,
      "loss": 1.0956,
      "step": 3509
    },
    {
      "epoch": 0.25300944280256615,
      "grad_norm": 0.12357313099752085,
      "learning_rate": 0.0001750401264627655,
      "loss": 1.0764,
      "step": 3510
    },
    {
      "epoch": 0.2530815252649031,
      "grad_norm": 0.21058245128942202,
      "learning_rate": 0.0001750246923580616,
      "loss": 1.0616,
      "step": 3511
    },
    {
      "epoch": 0.25315360772723994,
      "grad_norm": 0.1581993040520055,
      "learning_rate": 0.00017500925416384083,
      "loss": 1.1412,
      "step": 3512
    },
    {
      "epoch": 0.25322569018957686,
      "grad_norm": 0.16611571032631695,
      "learning_rate": 0.0001749938118809446,
      "loss": 1.091,
      "step": 3513
    },
    {
      "epoch": 0.2532977726519138,
      "grad_norm": 0.15301408303225875,
      "learning_rate": 0.00017497836551021474,
      "loss": 1.0406,
      "step": 3514
    },
    {
      "epoch": 0.2533698551142507,
      "grad_norm": 1.5359961876717514,
      "learning_rate": 0.0001749629150524932,
      "loss": 1.1367,
      "step": 3515
    },
    {
      "epoch": 0.2534419375765876,
      "grad_norm": 0.6105436089940409,
      "learning_rate": 0.0001749474605086221,
      "loss": 1.0632,
      "step": 3516
    },
    {
      "epoch": 0.25351402003892454,
      "grad_norm": 0.4613510855903628,
      "learning_rate": 0.00017493200187944393,
      "loss": 1.11,
      "step": 3517
    },
    {
      "epoch": 0.25358610250126146,
      "grad_norm": 0.1923018439039606,
      "learning_rate": 0.00017491653916580128,
      "loss": 0.9751,
      "step": 3518
    },
    {
      "epoch": 0.2536581849635984,
      "grad_norm": 0.15997610821868105,
      "learning_rate": 0.00017490107236853706,
      "loss": 1.0585,
      "step": 3519
    },
    {
      "epoch": 0.2537302674259353,
      "grad_norm": 0.12031227837541791,
      "learning_rate": 0.00017488560148849427,
      "loss": 1.0261,
      "step": 3520
    },
    {
      "epoch": 0.25380234988827216,
      "grad_norm": 0.12915172319108736,
      "learning_rate": 0.0001748701265265163,
      "loss": 1.0661,
      "step": 3521
    },
    {
      "epoch": 0.2538744323506091,
      "grad_norm": 0.21455506116135753,
      "learning_rate": 0.00017485464748344658,
      "loss": 1.1022,
      "step": 3522
    },
    {
      "epoch": 0.253946514812946,
      "grad_norm": 0.12242480873842644,
      "learning_rate": 0.00017483916436012893,
      "loss": 1.1211,
      "step": 3523
    },
    {
      "epoch": 0.2540185972752829,
      "grad_norm": 0.18319429008949017,
      "learning_rate": 0.00017482367715740727,
      "loss": 1.1174,
      "step": 3524
    },
    {
      "epoch": 0.25409067973761984,
      "grad_norm": 0.18612920526909602,
      "learning_rate": 0.00017480818587612584,
      "loss": 1.1203,
      "step": 3525
    },
    {
      "epoch": 0.25416276219995676,
      "grad_norm": 0.15269233895919881,
      "learning_rate": 0.00017479269051712902,
      "loss": 1.1113,
      "step": 3526
    },
    {
      "epoch": 0.2542348446622937,
      "grad_norm": 0.3465234573403923,
      "learning_rate": 0.00017477719108126145,
      "loss": 0.9632,
      "step": 3527
    },
    {
      "epoch": 0.2543069271246306,
      "grad_norm": 0.26411175909867207,
      "learning_rate": 0.00017476168756936798,
      "loss": 1.1023,
      "step": 3528
    },
    {
      "epoch": 0.25437900958696746,
      "grad_norm": 0.14428395684020442,
      "learning_rate": 0.00017474617998229373,
      "loss": 1.0504,
      "step": 3529
    },
    {
      "epoch": 0.2544510920493044,
      "grad_norm": 0.14589092028488593,
      "learning_rate": 0.00017473066832088393,
      "loss": 1.0941,
      "step": 3530
    },
    {
      "epoch": 0.2545231745116413,
      "grad_norm": 0.19841826833054296,
      "learning_rate": 0.00017471515258598417,
      "loss": 1.0528,
      "step": 3531
    },
    {
      "epoch": 0.2545952569739782,
      "grad_norm": 0.13616325057428408,
      "learning_rate": 0.00017469963277844016,
      "loss": 1.036,
      "step": 3532
    },
    {
      "epoch": 0.25466733943631514,
      "grad_norm": 0.1734949015705097,
      "learning_rate": 0.0001746841088990979,
      "loss": 1.0495,
      "step": 3533
    },
    {
      "epoch": 0.25473942189865206,
      "grad_norm": 0.21665313214530207,
      "learning_rate": 0.00017466858094880355,
      "loss": 1.0765,
      "step": 3534
    },
    {
      "epoch": 0.254811504360989,
      "grad_norm": 0.22696067451390015,
      "learning_rate": 0.00017465304892840355,
      "loss": 1.0087,
      "step": 3535
    },
    {
      "epoch": 0.2548835868233259,
      "grad_norm": 0.4318760718943441,
      "learning_rate": 0.0001746375128387445,
      "loss": 1.0821,
      "step": 3536
    },
    {
      "epoch": 0.2549556692856628,
      "grad_norm": 0.1352390499130228,
      "learning_rate": 0.00017462197268067328,
      "loss": 1.1041,
      "step": 3537
    },
    {
      "epoch": 0.2550277517479997,
      "grad_norm": 0.1203894755822425,
      "learning_rate": 0.00017460642845503697,
      "loss": 0.9948,
      "step": 3538
    },
    {
      "epoch": 0.2550998342103366,
      "grad_norm": 0.22503131910582946,
      "learning_rate": 0.00017459088016268286,
      "loss": 1.0394,
      "step": 3539
    },
    {
      "epoch": 0.2551719166726735,
      "grad_norm": 0.16279379920161163,
      "learning_rate": 0.00017457532780445844,
      "loss": 0.9734,
      "step": 3540
    },
    {
      "epoch": 0.25524399913501045,
      "grad_norm": 0.12016959207904651,
      "learning_rate": 0.00017455977138121153,
      "loss": 1.0522,
      "step": 3541
    },
    {
      "epoch": 0.25531608159734737,
      "grad_norm": 1.292390869737488,
      "learning_rate": 0.00017454421089379004,
      "loss": 1.085,
      "step": 3542
    },
    {
      "epoch": 0.2553881640596843,
      "grad_norm": 0.25285019253712854,
      "learning_rate": 0.00017452864634304212,
      "loss": 1.146,
      "step": 3543
    },
    {
      "epoch": 0.2554602465220212,
      "grad_norm": 0.28503205604486487,
      "learning_rate": 0.0001745130777298163,
      "loss": 1.1253,
      "step": 3544
    },
    {
      "epoch": 0.2555323289843581,
      "grad_norm": 0.6448447832507529,
      "learning_rate": 0.00017449750505496106,
      "loss": 1.0252,
      "step": 3545
    },
    {
      "epoch": 0.25560441144669505,
      "grad_norm": 0.11925793162797692,
      "learning_rate": 0.00017448192831932534,
      "loss": 1.1021,
      "step": 3546
    },
    {
      "epoch": 0.2556764939090319,
      "grad_norm": 0.3108280900967145,
      "learning_rate": 0.00017446634752375822,
      "loss": 1.1522,
      "step": 3547
    },
    {
      "epoch": 0.25574857637136883,
      "grad_norm": 0.38633111220425465,
      "learning_rate": 0.00017445076266910892,
      "loss": 1.0217,
      "step": 3548
    },
    {
      "epoch": 0.25582065883370575,
      "grad_norm": 0.21794240040722335,
      "learning_rate": 0.00017443517375622704,
      "loss": 1.038,
      "step": 3549
    },
    {
      "epoch": 0.25589274129604267,
      "grad_norm": 0.2687922357362014,
      "learning_rate": 0.00017441958078596225,
      "loss": 1.1412,
      "step": 3550
    },
    {
      "epoch": 0.2559648237583796,
      "grad_norm": 0.3045145339115393,
      "learning_rate": 0.00017440398375916455,
      "loss": 1.0909,
      "step": 3551
    },
    {
      "epoch": 0.2560369062207165,
      "grad_norm": 0.12354506763302321,
      "learning_rate": 0.00017438838267668408,
      "loss": 1.1199,
      "step": 3552
    },
    {
      "epoch": 0.25610898868305343,
      "grad_norm": 0.14063827025404527,
      "learning_rate": 0.00017437277753937127,
      "loss": 1.1451,
      "step": 3553
    },
    {
      "epoch": 0.25618107114539035,
      "grad_norm": 0.3368778547563234,
      "learning_rate": 0.00017435716834807673,
      "loss": 1.0636,
      "step": 3554
    },
    {
      "epoch": 0.2562531536077272,
      "grad_norm": 0.21409795892373187,
      "learning_rate": 0.00017434155510365125,
      "loss": 1.0747,
      "step": 3555
    },
    {
      "epoch": 0.25632523607006413,
      "grad_norm": 0.1287822498103368,
      "learning_rate": 0.000174325937806946,
      "loss": 1.0337,
      "step": 3556
    },
    {
      "epoch": 0.25639731853240105,
      "grad_norm": 0.14907710517559591,
      "learning_rate": 0.00017431031645881215,
      "loss": 1.0143,
      "step": 3557
    },
    {
      "epoch": 0.256469400994738,
      "grad_norm": 0.297283199803812,
      "learning_rate": 0.0001742946910601013,
      "loss": 1.1119,
      "step": 3558
    },
    {
      "epoch": 0.2565414834570749,
      "grad_norm": 0.23339359462718068,
      "learning_rate": 0.0001742790616116651,
      "loss": 1.1551,
      "step": 3559
    },
    {
      "epoch": 0.2566135659194118,
      "grad_norm": 0.16183256996348003,
      "learning_rate": 0.00017426342811435554,
      "loss": 1.1159,
      "step": 3560
    },
    {
      "epoch": 0.25668564838174873,
      "grad_norm": 0.12069391803264708,
      "learning_rate": 0.00017424779056902473,
      "loss": 1.0878,
      "step": 3561
    },
    {
      "epoch": 0.25675773084408565,
      "grad_norm": 0.11137006522967056,
      "learning_rate": 0.00017423214897652515,
      "loss": 1.0965,
      "step": 3562
    },
    {
      "epoch": 0.25682981330642257,
      "grad_norm": 0.18478072747204738,
      "learning_rate": 0.0001742165033377093,
      "loss": 1.172,
      "step": 3563
    },
    {
      "epoch": 0.25690189576875944,
      "grad_norm": 0.14811633621611264,
      "learning_rate": 0.0001742008536534301,
      "loss": 1.1035,
      "step": 3564
    },
    {
      "epoch": 0.25697397823109636,
      "grad_norm": 0.224722915683393,
      "learning_rate": 0.00017418519992454053,
      "loss": 1.0356,
      "step": 3565
    },
    {
      "epoch": 0.2570460606934333,
      "grad_norm": 0.1516458395107159,
      "learning_rate": 0.0001741695421518939,
      "loss": 1.0552,
      "step": 3566
    },
    {
      "epoch": 0.2571181431557702,
      "grad_norm": 0.132831891459912,
      "learning_rate": 0.00017415388033634364,
      "loss": 1.0682,
      "step": 3567
    },
    {
      "epoch": 0.2571902256181071,
      "grad_norm": 0.11405796623322977,
      "learning_rate": 0.00017413821447874355,
      "loss": 1.065,
      "step": 3568
    },
    {
      "epoch": 0.25726230808044404,
      "grad_norm": 0.1504495157639993,
      "learning_rate": 0.0001741225445799475,
      "loss": 1.0393,
      "step": 3569
    },
    {
      "epoch": 0.25733439054278096,
      "grad_norm": 0.5442042175821536,
      "learning_rate": 0.00017410687064080968,
      "loss": 1.1793,
      "step": 3570
    },
    {
      "epoch": 0.2574064730051179,
      "grad_norm": 0.13872054337575843,
      "learning_rate": 0.00017409119266218438,
      "loss": 1.1107,
      "step": 3571
    },
    {
      "epoch": 0.2574785554674548,
      "grad_norm": 0.2309577453548078,
      "learning_rate": 0.00017407551064492626,
      "loss": 1.1356,
      "step": 3572
    },
    {
      "epoch": 0.25755063792979166,
      "grad_norm": 0.19816275121332275,
      "learning_rate": 0.0001740598245898901,
      "loss": 1.1458,
      "step": 3573
    },
    {
      "epoch": 0.2576227203921286,
      "grad_norm": 0.16733842774184074,
      "learning_rate": 0.00017404413449793098,
      "loss": 1.1023,
      "step": 3574
    },
    {
      "epoch": 0.2576948028544655,
      "grad_norm": 0.1720696376431023,
      "learning_rate": 0.0001740284403699041,
      "loss": 1.0125,
      "step": 3575
    },
    {
      "epoch": 0.2577668853168024,
      "grad_norm": 0.3431018241120268,
      "learning_rate": 0.00017401274220666492,
      "loss": 1.1571,
      "step": 3576
    },
    {
      "epoch": 0.25783896777913934,
      "grad_norm": 0.10520551823407213,
      "learning_rate": 0.00017399704000906915,
      "loss": 0.9618,
      "step": 3577
    },
    {
      "epoch": 0.25791105024147626,
      "grad_norm": 0.11956786910383559,
      "learning_rate": 0.0001739813337779727,
      "loss": 1.1349,
      "step": 3578
    },
    {
      "epoch": 0.2579831327038132,
      "grad_norm": 0.27692059526877194,
      "learning_rate": 0.00017396562351423175,
      "loss": 1.0326,
      "step": 3579
    },
    {
      "epoch": 0.2580552151661501,
      "grad_norm": 0.15179361981588646,
      "learning_rate": 0.00017394990921870256,
      "loss": 1.038,
      "step": 3580
    },
    {
      "epoch": 0.25812729762848696,
      "grad_norm": 0.22626732140717867,
      "learning_rate": 0.00017393419089224176,
      "loss": 1.0692,
      "step": 3581
    },
    {
      "epoch": 0.2581993800908239,
      "grad_norm": 0.35611031831403495,
      "learning_rate": 0.00017391846853570612,
      "loss": 1.1064,
      "step": 3582
    },
    {
      "epoch": 0.2582714625531608,
      "grad_norm": 0.12409381070088772,
      "learning_rate": 0.00017390274214995263,
      "loss": 1.0134,
      "step": 3583
    },
    {
      "epoch": 0.2583435450154977,
      "grad_norm": 0.1617141798574501,
      "learning_rate": 0.00017388701173583858,
      "loss": 1.1534,
      "step": 3584
    },
    {
      "epoch": 0.25841562747783464,
      "grad_norm": 0.14223208370005877,
      "learning_rate": 0.00017387127729422133,
      "loss": 1.0247,
      "step": 3585
    },
    {
      "epoch": 0.25848770994017156,
      "grad_norm": 0.13574493577911792,
      "learning_rate": 0.00017385553882595862,
      "loss": 1.0367,
      "step": 3586
    },
    {
      "epoch": 0.2585597924025085,
      "grad_norm": 0.2400659307975102,
      "learning_rate": 0.00017383979633190833,
      "loss": 1.045,
      "step": 3587
    },
    {
      "epoch": 0.2586318748648454,
      "grad_norm": 0.32636173602397134,
      "learning_rate": 0.00017382404981292852,
      "loss": 1.0726,
      "step": 3588
    },
    {
      "epoch": 0.2587039573271823,
      "grad_norm": 0.12066600632351314,
      "learning_rate": 0.00017380829926987756,
      "loss": 1.0961,
      "step": 3589
    },
    {
      "epoch": 0.2587760397895192,
      "grad_norm": 0.266207843266545,
      "learning_rate": 0.00017379254470361398,
      "loss": 1.0903,
      "step": 3590
    },
    {
      "epoch": 0.2588481222518561,
      "grad_norm": 0.145292787982517,
      "learning_rate": 0.00017377678611499658,
      "loss": 1.0777,
      "step": 3591
    },
    {
      "epoch": 0.258920204714193,
      "grad_norm": 0.10900965636577846,
      "learning_rate": 0.00017376102350488427,
      "loss": 1.0315,
      "step": 3592
    },
    {
      "epoch": 0.25899228717652995,
      "grad_norm": 0.1220814766375602,
      "learning_rate": 0.00017374525687413633,
      "loss": 1.1547,
      "step": 3593
    },
    {
      "epoch": 0.25906436963886686,
      "grad_norm": 0.25123269377334057,
      "learning_rate": 0.00017372948622361212,
      "loss": 1.1627,
      "step": 3594
    },
    {
      "epoch": 0.2591364521012038,
      "grad_norm": 0.14521741500919128,
      "learning_rate": 0.00017371371155417128,
      "loss": 1.0819,
      "step": 3595
    },
    {
      "epoch": 0.2592085345635407,
      "grad_norm": 0.1446697102499418,
      "learning_rate": 0.00017369793286667372,
      "loss": 1.0219,
      "step": 3596
    },
    {
      "epoch": 0.2592806170258776,
      "grad_norm": 0.16888177603009527,
      "learning_rate": 0.0001736821501619795,
      "loss": 1.0923,
      "step": 3597
    },
    {
      "epoch": 0.25935269948821454,
      "grad_norm": 0.18485458148457287,
      "learning_rate": 0.00017366636344094894,
      "loss": 1.0773,
      "step": 3598
    },
    {
      "epoch": 0.2594247819505514,
      "grad_norm": 0.2110645118589822,
      "learning_rate": 0.00017365057270444254,
      "loss": 1.1298,
      "step": 3599
    },
    {
      "epoch": 0.25949686441288833,
      "grad_norm": 0.5605044255241916,
      "learning_rate": 0.000173634777953321,
      "loss": 1.1614,
      "step": 3600
    },
    {
      "epoch": 0.25956894687522525,
      "grad_norm": 0.1423696102406303,
      "learning_rate": 0.0001736189791884453,
      "loss": 1.1289,
      "step": 3601
    },
    {
      "epoch": 0.25964102933756217,
      "grad_norm": 1.173373627581152,
      "learning_rate": 0.00017360317641067666,
      "loss": 1.125,
      "step": 3602
    },
    {
      "epoch": 0.2597131117998991,
      "grad_norm": 0.17394098175376937,
      "learning_rate": 0.0001735873696208764,
      "loss": 1.011,
      "step": 3603
    },
    {
      "epoch": 0.259785194262236,
      "grad_norm": 0.12754726746027603,
      "learning_rate": 0.00017357155881990617,
      "loss": 1.0581,
      "step": 3604
    },
    {
      "epoch": 0.2598572767245729,
      "grad_norm": 0.15506432452230615,
      "learning_rate": 0.00017355574400862778,
      "loss": 1.0555,
      "step": 3605
    },
    {
      "epoch": 0.25992935918690985,
      "grad_norm": 0.38501249761775774,
      "learning_rate": 0.00017353992518790332,
      "loss": 1.0215,
      "step": 3606
    },
    {
      "epoch": 0.2600014416492467,
      "grad_norm": 0.14088259529141353,
      "learning_rate": 0.00017352410235859503,
      "loss": 1.0373,
      "step": 3607
    },
    {
      "epoch": 0.26007352411158363,
      "grad_norm": 0.15719246524143984,
      "learning_rate": 0.0001735082755215654,
      "loss": 1.2217,
      "step": 3608
    },
    {
      "epoch": 0.26014560657392055,
      "grad_norm": 0.12851661144530377,
      "learning_rate": 0.00017349244467767708,
      "loss": 1.0633,
      "step": 3609
    },
    {
      "epoch": 0.26021768903625747,
      "grad_norm": 0.130882278719881,
      "learning_rate": 0.00017347660982779307,
      "loss": 1.0582,
      "step": 3610
    },
    {
      "epoch": 0.2602897714985944,
      "grad_norm": 0.21346482735722855,
      "learning_rate": 0.00017346077097277652,
      "loss": 1.0735,
      "step": 3611
    },
    {
      "epoch": 0.2603618539609313,
      "grad_norm": 0.2765572790548,
      "learning_rate": 0.0001734449281134907,
      "loss": 1.0762,
      "step": 3612
    },
    {
      "epoch": 0.26043393642326823,
      "grad_norm": 0.20050509609401765,
      "learning_rate": 0.00017342908125079928,
      "loss": 1.0843,
      "step": 3613
    },
    {
      "epoch": 0.26050601888560515,
      "grad_norm": 0.1520533342810812,
      "learning_rate": 0.000173413230385566,
      "loss": 1.0823,
      "step": 3614
    },
    {
      "epoch": 0.26057810134794207,
      "grad_norm": 0.345561349815445,
      "learning_rate": 0.0001733973755186549,
      "loss": 1.107,
      "step": 3615
    },
    {
      "epoch": 0.26065018381027893,
      "grad_norm": 0.16280467531618953,
      "learning_rate": 0.0001733815166509302,
      "loss": 1.0988,
      "step": 3616
    },
    {
      "epoch": 0.26072226627261585,
      "grad_norm": 0.12329368477478557,
      "learning_rate": 0.00017336565378325632,
      "loss": 1.1554,
      "step": 3617
    },
    {
      "epoch": 0.2607943487349528,
      "grad_norm": 0.253571196954446,
      "learning_rate": 0.00017334978691649802,
      "loss": 1.1074,
      "step": 3618
    },
    {
      "epoch": 0.2608664311972897,
      "grad_norm": 0.14571973981201222,
      "learning_rate": 0.00017333391605152008,
      "loss": 1.0509,
      "step": 3619
    },
    {
      "epoch": 0.2609385136596266,
      "grad_norm": 0.12523018536863254,
      "learning_rate": 0.00017331804118918766,
      "loss": 1.0399,
      "step": 3620
    },
    {
      "epoch": 0.26101059612196353,
      "grad_norm": 0.10815468799499273,
      "learning_rate": 0.00017330216233036612,
      "loss": 1.0481,
      "step": 3621
    },
    {
      "epoch": 0.26108267858430045,
      "grad_norm": 0.14469012760296618,
      "learning_rate": 0.00017328627947592087,
      "loss": 1.0975,
      "step": 3622
    },
    {
      "epoch": 0.2611547610466374,
      "grad_norm": 0.2535519608892617,
      "learning_rate": 0.0001732703926267178,
      "loss": 1.0958,
      "step": 3623
    },
    {
      "epoch": 0.2612268435089743,
      "grad_norm": 0.1312108343457039,
      "learning_rate": 0.00017325450178362286,
      "loss": 1.0697,
      "step": 3624
    },
    {
      "epoch": 0.26129892597131116,
      "grad_norm": 0.12624788152035737,
      "learning_rate": 0.00017323860694750217,
      "loss": 1.049,
      "step": 3625
    },
    {
      "epoch": 0.2613710084336481,
      "grad_norm": 0.1739296609374371,
      "learning_rate": 0.00017322270811922222,
      "loss": 1.0474,
      "step": 3626
    },
    {
      "epoch": 0.261443090895985,
      "grad_norm": 0.25899714977646177,
      "learning_rate": 0.0001732068052996496,
      "loss": 1.141,
      "step": 3627
    },
    {
      "epoch": 0.2615151733583219,
      "grad_norm": 0.25437731049004475,
      "learning_rate": 0.00017319089848965118,
      "loss": 1.0522,
      "step": 3628
    },
    {
      "epoch": 0.26158725582065884,
      "grad_norm": 0.10716461328854007,
      "learning_rate": 0.00017317498769009397,
      "loss": 1.1151,
      "step": 3629
    },
    {
      "epoch": 0.26165933828299576,
      "grad_norm": 0.1549815252997683,
      "learning_rate": 0.0001731590729018453,
      "loss": 1.1456,
      "step": 3630
    },
    {
      "epoch": 0.2617314207453327,
      "grad_norm": 0.17890359656209473,
      "learning_rate": 0.0001731431541257727,
      "loss": 1.1587,
      "step": 3631
    },
    {
      "epoch": 0.2618035032076696,
      "grad_norm": 0.30321432239225643,
      "learning_rate": 0.0001731272313627438,
      "loss": 1.1264,
      "step": 3632
    },
    {
      "epoch": 0.26187558567000646,
      "grad_norm": 0.19734400354905995,
      "learning_rate": 0.00017311130461362658,
      "loss": 0.9726,
      "step": 3633
    },
    {
      "epoch": 0.2619476681323434,
      "grad_norm": 0.2488147460944119,
      "learning_rate": 0.0001730953738792892,
      "loss": 1.1766,
      "step": 3634
    },
    {
      "epoch": 0.2620197505946803,
      "grad_norm": 0.13920514345569246,
      "learning_rate": 0.0001730794391606,
      "loss": 1.0471,
      "step": 3635
    },
    {
      "epoch": 0.2620918330570172,
      "grad_norm": 0.10905485934858833,
      "learning_rate": 0.0001730635004584276,
      "loss": 1.1392,
      "step": 3636
    },
    {
      "epoch": 0.26216391551935414,
      "grad_norm": 0.16325175569873443,
      "learning_rate": 0.00017304755777364074,
      "loss": 1.0974,
      "step": 3637
    },
    {
      "epoch": 0.26223599798169106,
      "grad_norm": 0.1344526244674064,
      "learning_rate": 0.0001730316111071085,
      "loss": 1.0806,
      "step": 3638
    },
    {
      "epoch": 0.262308080444028,
      "grad_norm": 0.13825639731440653,
      "learning_rate": 0.0001730156604597001,
      "loss": 1.0615,
      "step": 3639
    },
    {
      "epoch": 0.2623801629063649,
      "grad_norm": 0.24500913862961338,
      "learning_rate": 0.000172999705832285,
      "loss": 1.1385,
      "step": 3640
    },
    {
      "epoch": 0.2624522453687018,
      "grad_norm": 0.34531127376957627,
      "learning_rate": 0.00017298374722573282,
      "loss": 1.1458,
      "step": 3641
    },
    {
      "epoch": 0.2625243278310387,
      "grad_norm": 0.1722868876059448,
      "learning_rate": 0.00017296778464091352,
      "loss": 1.0992,
      "step": 3642
    },
    {
      "epoch": 0.2625964102933756,
      "grad_norm": 0.11228058707294584,
      "learning_rate": 0.00017295181807869717,
      "loss": 1.0551,
      "step": 3643
    },
    {
      "epoch": 0.2626684927557125,
      "grad_norm": 0.13023977856939428,
      "learning_rate": 0.00017293584753995405,
      "loss": 1.1239,
      "step": 3644
    },
    {
      "epoch": 0.26274057521804944,
      "grad_norm": 0.1429839542499905,
      "learning_rate": 0.0001729198730255548,
      "loss": 1.0898,
      "step": 3645
    },
    {
      "epoch": 0.26281265768038636,
      "grad_norm": 0.9404828866381865,
      "learning_rate": 0.00017290389453637005,
      "loss": 1.085,
      "step": 3646
    },
    {
      "epoch": 0.2628847401427233,
      "grad_norm": 0.16639948986363423,
      "learning_rate": 0.00017288791207327087,
      "loss": 1.0783,
      "step": 3647
    },
    {
      "epoch": 0.2629568226050602,
      "grad_norm": 0.11993307489372912,
      "learning_rate": 0.0001728719256371284,
      "loss": 1.1318,
      "step": 3648
    },
    {
      "epoch": 0.2630289050673971,
      "grad_norm": 0.12829980263662188,
      "learning_rate": 0.000172855935228814,
      "loss": 1.0869,
      "step": 3649
    },
    {
      "epoch": 0.26310098752973404,
      "grad_norm": 0.29925779314391693,
      "learning_rate": 0.0001728399408491994,
      "loss": 1.116,
      "step": 3650
    },
    {
      "epoch": 0.2631730699920709,
      "grad_norm": 0.10273870038720276,
      "learning_rate": 0.0001728239424991564,
      "loss": 1.0282,
      "step": 3651
    },
    {
      "epoch": 0.2632451524544078,
      "grad_norm": 0.1386163614008476,
      "learning_rate": 0.00017280794017955698,
      "loss": 1.0789,
      "step": 3652
    },
    {
      "epoch": 0.26331723491674475,
      "grad_norm": 0.13722546884926276,
      "learning_rate": 0.00017279193389127347,
      "loss": 1.0655,
      "step": 3653
    },
    {
      "epoch": 0.26338931737908167,
      "grad_norm": 0.15933094488842262,
      "learning_rate": 0.0001727759236351784,
      "loss": 1.0841,
      "step": 3654
    },
    {
      "epoch": 0.2634613998414186,
      "grad_norm": 0.21052271330155684,
      "learning_rate": 0.00017275990941214436,
      "loss": 1.1349,
      "step": 3655
    },
    {
      "epoch": 0.2635334823037555,
      "grad_norm": 0.49368158625947006,
      "learning_rate": 0.00017274389122304436,
      "loss": 1.1379,
      "step": 3656
    },
    {
      "epoch": 0.2636055647660924,
      "grad_norm": 0.1574585151899983,
      "learning_rate": 0.00017272786906875147,
      "loss": 1.069,
      "step": 3657
    },
    {
      "epoch": 0.26367764722842935,
      "grad_norm": 0.12659476774729717,
      "learning_rate": 0.0001727118429501391,
      "loss": 1.0745,
      "step": 3658
    },
    {
      "epoch": 0.2637497296907662,
      "grad_norm": 0.1366834831066047,
      "learning_rate": 0.00017269581286808076,
      "loss": 1.0513,
      "step": 3659
    },
    {
      "epoch": 0.26382181215310313,
      "grad_norm": 0.19248112666470302,
      "learning_rate": 0.0001726797788234503,
      "loss": 1.0577,
      "step": 3660
    },
    {
      "epoch": 0.26389389461544005,
      "grad_norm": 0.16463159793875504,
      "learning_rate": 0.00017266374081712166,
      "loss": 1.1325,
      "step": 3661
    },
    {
      "epoch": 0.26396597707777697,
      "grad_norm": 0.19975488631506702,
      "learning_rate": 0.0001726476988499691,
      "loss": 1.0641,
      "step": 3662
    },
    {
      "epoch": 0.2640380595401139,
      "grad_norm": 0.16212416286222273,
      "learning_rate": 0.000172631652922867,
      "loss": 1.108,
      "step": 3663
    },
    {
      "epoch": 0.2641101420024508,
      "grad_norm": 0.14456533098821892,
      "learning_rate": 0.00017261560303669007,
      "loss": 1.0959,
      "step": 3664
    },
    {
      "epoch": 0.26418222446478773,
      "grad_norm": 0.2066448641798533,
      "learning_rate": 0.0001725995491923131,
      "loss": 1.11,
      "step": 3665
    },
    {
      "epoch": 0.26425430692712465,
      "grad_norm": 0.1927636211559309,
      "learning_rate": 0.0001725834913906112,
      "loss": 1.1274,
      "step": 3666
    },
    {
      "epoch": 0.26432638938946157,
      "grad_norm": 0.1696491632637756,
      "learning_rate": 0.0001725674296324597,
      "loss": 1.0634,
      "step": 3667
    },
    {
      "epoch": 0.26439847185179843,
      "grad_norm": 0.1197650026441018,
      "learning_rate": 0.00017255136391873408,
      "loss": 1.0069,
      "step": 3668
    },
    {
      "epoch": 0.26447055431413535,
      "grad_norm": 0.1703394818433544,
      "learning_rate": 0.00017253529425031005,
      "loss": 1.1094,
      "step": 3669
    },
    {
      "epoch": 0.2645426367764723,
      "grad_norm": 0.1430894171302283,
      "learning_rate": 0.00017251922062806355,
      "loss": 1.0746,
      "step": 3670
    },
    {
      "epoch": 0.2646147192388092,
      "grad_norm": 0.12305315408552496,
      "learning_rate": 0.00017250314305287078,
      "loss": 0.9898,
      "step": 3671
    },
    {
      "epoch": 0.2646868017011461,
      "grad_norm": 0.17764491625980164,
      "learning_rate": 0.00017248706152560804,
      "loss": 1.0045,
      "step": 3672
    },
    {
      "epoch": 0.26475888416348303,
      "grad_norm": 0.21170407405779818,
      "learning_rate": 0.000172470976047152,
      "loss": 1.1044,
      "step": 3673
    },
    {
      "epoch": 0.26483096662581995,
      "grad_norm": 0.25301442284247594,
      "learning_rate": 0.0001724548866183794,
      "loss": 1.1434,
      "step": 3674
    },
    {
      "epoch": 0.26490304908815687,
      "grad_norm": 0.49554268669331375,
      "learning_rate": 0.0001724387932401673,
      "loss": 1.1077,
      "step": 3675
    },
    {
      "epoch": 0.2649751315504938,
      "grad_norm": 0.15431734852136297,
      "learning_rate": 0.0001724226959133929,
      "loss": 1.0324,
      "step": 3676
    },
    {
      "epoch": 0.26504721401283066,
      "grad_norm": 0.18608461753445477,
      "learning_rate": 0.00017240659463893364,
      "loss": 1.032,
      "step": 3677
    },
    {
      "epoch": 0.2651192964751676,
      "grad_norm": 0.21624238845941746,
      "learning_rate": 0.00017239048941766725,
      "loss": 1.0991,
      "step": 3678
    },
    {
      "epoch": 0.2651913789375045,
      "grad_norm": 0.13514778460737248,
      "learning_rate": 0.00017237438025047153,
      "loss": 1.0225,
      "step": 3679
    },
    {
      "epoch": 0.2652634613998414,
      "grad_norm": 0.15686356445664418,
      "learning_rate": 0.0001723582671382246,
      "loss": 1.0668,
      "step": 3680
    },
    {
      "epoch": 0.26533554386217834,
      "grad_norm": 0.42191409330484053,
      "learning_rate": 0.00017234215008180478,
      "loss": 1.1285,
      "step": 3681
    },
    {
      "epoch": 0.26540762632451526,
      "grad_norm": 0.18001632380867008,
      "learning_rate": 0.00017232602908209062,
      "loss": 1.0126,
      "step": 3682
    },
    {
      "epoch": 0.2654797087868522,
      "grad_norm": 0.17902686015331692,
      "learning_rate": 0.0001723099041399608,
      "loss": 0.9725,
      "step": 3683
    },
    {
      "epoch": 0.2655517912491891,
      "grad_norm": 0.23944783351417187,
      "learning_rate": 0.0001722937752562943,
      "loss": 1.0569,
      "step": 3684
    },
    {
      "epoch": 0.26562387371152596,
      "grad_norm": 0.17108822112886649,
      "learning_rate": 0.0001722776424319703,
      "loss": 1.0491,
      "step": 3685
    },
    {
      "epoch": 0.2656959561738629,
      "grad_norm": 0.11336291164072948,
      "learning_rate": 0.00017226150566786817,
      "loss": 1.0868,
      "step": 3686
    },
    {
      "epoch": 0.2657680386361998,
      "grad_norm": 0.17689261133258244,
      "learning_rate": 0.0001722453649648675,
      "loss": 0.9577,
      "step": 3687
    },
    {
      "epoch": 0.2658401210985367,
      "grad_norm": 0.13469493179958447,
      "learning_rate": 0.00017222922032384812,
      "loss": 1.1121,
      "step": 3688
    },
    {
      "epoch": 0.26591220356087364,
      "grad_norm": 0.13063128554783576,
      "learning_rate": 0.0001722130717456901,
      "loss": 1.1133,
      "step": 3689
    },
    {
      "epoch": 0.26598428602321056,
      "grad_norm": 0.19171200682384842,
      "learning_rate": 0.00017219691923127357,
      "loss": 1.1088,
      "step": 3690
    },
    {
      "epoch": 0.2660563684855475,
      "grad_norm": 0.16584896120883144,
      "learning_rate": 0.00017218076278147906,
      "loss": 1.0271,
      "step": 3691
    },
    {
      "epoch": 0.2661284509478844,
      "grad_norm": 0.19582764865156085,
      "learning_rate": 0.00017216460239718724,
      "loss": 1.038,
      "step": 3692
    },
    {
      "epoch": 0.2662005334102213,
      "grad_norm": 0.12612599675333003,
      "learning_rate": 0.000172148438079279,
      "loss": 1.022,
      "step": 3693
    },
    {
      "epoch": 0.2662726158725582,
      "grad_norm": 0.1164783245282058,
      "learning_rate": 0.0001721322698286354,
      "loss": 1.1042,
      "step": 3694
    },
    {
      "epoch": 0.2663446983348951,
      "grad_norm": 0.19588459771478198,
      "learning_rate": 0.00017211609764613777,
      "loss": 1.0338,
      "step": 3695
    },
    {
      "epoch": 0.266416780797232,
      "grad_norm": 0.1492591786585441,
      "learning_rate": 0.0001720999215326677,
      "loss": 1.1212,
      "step": 3696
    },
    {
      "epoch": 0.26648886325956894,
      "grad_norm": 0.3108908607319386,
      "learning_rate": 0.00017208374148910686,
      "loss": 1.2016,
      "step": 3697
    },
    {
      "epoch": 0.26656094572190586,
      "grad_norm": 0.15689591566618635,
      "learning_rate": 0.00017206755751633722,
      "loss": 1.055,
      "step": 3698
    },
    {
      "epoch": 0.2666330281842428,
      "grad_norm": 0.15726991221676237,
      "learning_rate": 0.00017205136961524094,
      "loss": 0.9705,
      "step": 3699
    },
    {
      "epoch": 0.2667051106465797,
      "grad_norm": 0.22003056615368882,
      "learning_rate": 0.00017203517778670047,
      "loss": 1.0134,
      "step": 3700
    },
    {
      "epoch": 0.2667771931089166,
      "grad_norm": 0.1294702606094054,
      "learning_rate": 0.00017201898203159837,
      "loss": 1.1318,
      "step": 3701
    },
    {
      "epoch": 0.26684927557125354,
      "grad_norm": 0.15405430377345863,
      "learning_rate": 0.00017200278235081741,
      "loss": 0.982,
      "step": 3702
    },
    {
      "epoch": 0.2669213580335904,
      "grad_norm": 0.18488885480975778,
      "learning_rate": 0.00017198657874524067,
      "loss": 1.036,
      "step": 3703
    },
    {
      "epoch": 0.2669934404959273,
      "grad_norm": 0.11816474283916915,
      "learning_rate": 0.00017197037121575138,
      "loss": 1.1067,
      "step": 3704
    },
    {
      "epoch": 0.26706552295826425,
      "grad_norm": 0.1262457464484586,
      "learning_rate": 0.00017195415976323298,
      "loss": 1.0274,
      "step": 3705
    },
    {
      "epoch": 0.26713760542060117,
      "grad_norm": 0.11416983217099572,
      "learning_rate": 0.00017193794438856918,
      "loss": 1.0382,
      "step": 3706
    },
    {
      "epoch": 0.2672096878829381,
      "grad_norm": 0.12023278818735608,
      "learning_rate": 0.00017192172509264384,
      "loss": 1.1042,
      "step": 3707
    },
    {
      "epoch": 0.267281770345275,
      "grad_norm": 0.2667556513000076,
      "learning_rate": 0.000171905501876341,
      "loss": 1.0806,
      "step": 3708
    },
    {
      "epoch": 0.2673538528076119,
      "grad_norm": 0.15362434877746772,
      "learning_rate": 0.0001718892747405451,
      "loss": 0.9916,
      "step": 3709
    },
    {
      "epoch": 0.26742593526994884,
      "grad_norm": 0.17827974915121597,
      "learning_rate": 0.00017187304368614052,
      "loss": 1.166,
      "step": 3710
    },
    {
      "epoch": 0.2674980177322857,
      "grad_norm": 0.13086169021693533,
      "learning_rate": 0.0001718568087140121,
      "loss": 1.0459,
      "step": 3711
    },
    {
      "epoch": 0.26757010019462263,
      "grad_norm": 0.13747792018045998,
      "learning_rate": 0.00017184056982504472,
      "loss": 1.094,
      "step": 3712
    },
    {
      "epoch": 0.26764218265695955,
      "grad_norm": 0.1758164894372421,
      "learning_rate": 0.00017182432702012363,
      "loss": 1.0762,
      "step": 3713
    },
    {
      "epoch": 0.26771426511929647,
      "grad_norm": 0.11460027134809801,
      "learning_rate": 0.0001718080803001341,
      "loss": 1.0677,
      "step": 3714
    },
    {
      "epoch": 0.2677863475816334,
      "grad_norm": 0.12254777715697271,
      "learning_rate": 0.00017179182966596182,
      "loss": 1.1199,
      "step": 3715
    },
    {
      "epoch": 0.2678584300439703,
      "grad_norm": 0.2687540652173774,
      "learning_rate": 0.00017177557511849256,
      "loss": 1.0439,
      "step": 3716
    },
    {
      "epoch": 0.26793051250630723,
      "grad_norm": 0.11650842654789975,
      "learning_rate": 0.00017175931665861232,
      "loss": 1.1364,
      "step": 3717
    },
    {
      "epoch": 0.26800259496864415,
      "grad_norm": 0.13480877529926957,
      "learning_rate": 0.00017174305428720735,
      "loss": 1.0706,
      "step": 3718
    },
    {
      "epoch": 0.26807467743098107,
      "grad_norm": 0.12897025629747458,
      "learning_rate": 0.00017172678800516408,
      "loss": 0.9781,
      "step": 3719
    },
    {
      "epoch": 0.26814675989331793,
      "grad_norm": 0.15071278475041813,
      "learning_rate": 0.0001717105178133692,
      "loss": 1.0025,
      "step": 3720
    },
    {
      "epoch": 0.26821884235565485,
      "grad_norm": 0.26061623506418125,
      "learning_rate": 0.00017169424371270956,
      "loss": 1.0493,
      "step": 3721
    },
    {
      "epoch": 0.26829092481799177,
      "grad_norm": 0.12532490294749019,
      "learning_rate": 0.0001716779657040722,
      "loss": 1.0851,
      "step": 3722
    },
    {
      "epoch": 0.2683630072803287,
      "grad_norm": 0.14330122282107363,
      "learning_rate": 0.00017166168378834448,
      "loss": 1.0873,
      "step": 3723
    },
    {
      "epoch": 0.2684350897426656,
      "grad_norm": 0.19770409758932375,
      "learning_rate": 0.0001716453979664139,
      "loss": 1.081,
      "step": 3724
    },
    {
      "epoch": 0.26850717220500253,
      "grad_norm": 0.13069040085830838,
      "learning_rate": 0.00017162910823916816,
      "loss": 1.0146,
      "step": 3725
    },
    {
      "epoch": 0.26857925466733945,
      "grad_norm": 0.10921860178756722,
      "learning_rate": 0.0001716128146074952,
      "loss": 1.0384,
      "step": 3726
    },
    {
      "epoch": 0.26865133712967637,
      "grad_norm": 0.12733667002193352,
      "learning_rate": 0.00017159651707228323,
      "loss": 1.0895,
      "step": 3727
    },
    {
      "epoch": 0.2687234195920133,
      "grad_norm": 0.1575328173882526,
      "learning_rate": 0.0001715802156344205,
      "loss": 1.0989,
      "step": 3728
    },
    {
      "epoch": 0.26879550205435015,
      "grad_norm": 0.20389316563713397,
      "learning_rate": 0.0001715639102947957,
      "loss": 1.0547,
      "step": 3729
    },
    {
      "epoch": 0.2688675845166871,
      "grad_norm": 0.10770407194744235,
      "learning_rate": 0.00017154760105429748,
      "loss": 1.0615,
      "step": 3730
    },
    {
      "epoch": 0.268939666979024,
      "grad_norm": 0.12164785385900521,
      "learning_rate": 0.000171531287913815,
      "loss": 1.0373,
      "step": 3731
    },
    {
      "epoch": 0.2690117494413609,
      "grad_norm": 0.12186274041810588,
      "learning_rate": 0.00017151497087423734,
      "loss": 1.0631,
      "step": 3732
    },
    {
      "epoch": 0.26908383190369783,
      "grad_norm": 0.14140157733961461,
      "learning_rate": 0.00017149864993645398,
      "loss": 1.0825,
      "step": 3733
    },
    {
      "epoch": 0.26915591436603475,
      "grad_norm": 0.12480562870341566,
      "learning_rate": 0.0001714823251013546,
      "loss": 1.0766,
      "step": 3734
    },
    {
      "epoch": 0.2692279968283717,
      "grad_norm": 0.1260380317540183,
      "learning_rate": 0.00017146599636982894,
      "loss": 1.0896,
      "step": 3735
    },
    {
      "epoch": 0.2693000792907086,
      "grad_norm": 0.1380801917588273,
      "learning_rate": 0.00017144966374276716,
      "loss": 1.0607,
      "step": 3736
    },
    {
      "epoch": 0.26937216175304546,
      "grad_norm": 0.22029552020676532,
      "learning_rate": 0.0001714333272210595,
      "loss": 1.0139,
      "step": 3737
    },
    {
      "epoch": 0.2694442442153824,
      "grad_norm": 0.1229432047021212,
      "learning_rate": 0.0001714169868055964,
      "loss": 1.0793,
      "step": 3738
    },
    {
      "epoch": 0.2695163266777193,
      "grad_norm": 0.1513213191798443,
      "learning_rate": 0.00017140064249726865,
      "loss": 1.0852,
      "step": 3739
    },
    {
      "epoch": 0.2695884091400562,
      "grad_norm": 0.12660099024604976,
      "learning_rate": 0.00017138429429696712,
      "loss": 1.056,
      "step": 3740
    },
    {
      "epoch": 0.26966049160239314,
      "grad_norm": 0.14131476994836942,
      "learning_rate": 0.0001713679422055829,
      "loss": 1.0167,
      "step": 3741
    },
    {
      "epoch": 0.26973257406473006,
      "grad_norm": 0.11643450872033961,
      "learning_rate": 0.00017135158622400739,
      "loss": 1.1327,
      "step": 3742
    },
    {
      "epoch": 0.269804656527067,
      "grad_norm": 0.7169678616862822,
      "learning_rate": 0.0001713352263531321,
      "loss": 0.9648,
      "step": 3743
    },
    {
      "epoch": 0.2698767389894039,
      "grad_norm": 0.32000599239827837,
      "learning_rate": 0.00017131886259384875,
      "loss": 1.1844,
      "step": 3744
    },
    {
      "epoch": 0.2699488214517408,
      "grad_norm": 0.11540561336413162,
      "learning_rate": 0.00017130249494704938,
      "loss": 0.9822,
      "step": 3745
    },
    {
      "epoch": 0.2700209039140777,
      "grad_norm": 0.14547451595033453,
      "learning_rate": 0.00017128612341362612,
      "loss": 1.1063,
      "step": 3746
    },
    {
      "epoch": 0.2700929863764146,
      "grad_norm": 0.12605092651143357,
      "learning_rate": 0.00017126974799447142,
      "loss": 1.0169,
      "step": 3747
    },
    {
      "epoch": 0.2701650688387515,
      "grad_norm": 0.1374589834867198,
      "learning_rate": 0.00017125336869047787,
      "loss": 1.188,
      "step": 3748
    },
    {
      "epoch": 0.27023715130108844,
      "grad_norm": 0.13403529001988312,
      "learning_rate": 0.00017123698550253823,
      "loss": 1.1075,
      "step": 3749
    },
    {
      "epoch": 0.27030923376342536,
      "grad_norm": 0.26930363812635244,
      "learning_rate": 0.0001712205984315456,
      "loss": 1.1215,
      "step": 3750
    },
    {
      "epoch": 0.2703813162257623,
      "grad_norm": 0.11658750593339359,
      "learning_rate": 0.00017120420747839319,
      "loss": 1.0977,
      "step": 3751
    },
    {
      "epoch": 0.2704533986880992,
      "grad_norm": 0.1336325045785843,
      "learning_rate": 0.00017118781264397446,
      "loss": 1.0686,
      "step": 3752
    },
    {
      "epoch": 0.2705254811504361,
      "grad_norm": 0.10945325056907285,
      "learning_rate": 0.00017117141392918308,
      "loss": 1.0335,
      "step": 3753
    },
    {
      "epoch": 0.27059756361277304,
      "grad_norm": 0.3239666735077279,
      "learning_rate": 0.00017115501133491294,
      "loss": 1.0552,
      "step": 3754
    },
    {
      "epoch": 0.2706696460751099,
      "grad_norm": 0.11586518338515492,
      "learning_rate": 0.0001711386048620581,
      "loss": 1.0391,
      "step": 3755
    },
    {
      "epoch": 0.2707417285374468,
      "grad_norm": 0.15574409723942378,
      "learning_rate": 0.00017112219451151286,
      "loss": 1.0812,
      "step": 3756
    },
    {
      "epoch": 0.27081381099978374,
      "grad_norm": 0.1483951177672903,
      "learning_rate": 0.00017110578028417176,
      "loss": 0.9899,
      "step": 3757
    },
    {
      "epoch": 0.27088589346212066,
      "grad_norm": 0.13848915046375251,
      "learning_rate": 0.00017108936218092948,
      "loss": 1.095,
      "step": 3758
    },
    {
      "epoch": 0.2709579759244576,
      "grad_norm": 0.22144306303217653,
      "learning_rate": 0.000171072940202681,
      "loss": 1.1259,
      "step": 3759
    },
    {
      "epoch": 0.2710300583867945,
      "grad_norm": 0.1301644402882596,
      "learning_rate": 0.00017105651435032142,
      "loss": 1.0535,
      "step": 3760
    },
    {
      "epoch": 0.2711021408491314,
      "grad_norm": 0.12039312504967593,
      "learning_rate": 0.00017104008462474614,
      "loss": 1.0467,
      "step": 3761
    },
    {
      "epoch": 0.27117422331146834,
      "grad_norm": 0.1598211635202458,
      "learning_rate": 0.0001710236510268507,
      "loss": 1.042,
      "step": 3762
    },
    {
      "epoch": 0.2712463057738052,
      "grad_norm": 0.14181515306618644,
      "learning_rate": 0.00017100721355753086,
      "loss": 1.097,
      "step": 3763
    },
    {
      "epoch": 0.2713183882361421,
      "grad_norm": 0.21115096385969534,
      "learning_rate": 0.00017099077221768266,
      "loss": 1.0434,
      "step": 3764
    },
    {
      "epoch": 0.27139047069847905,
      "grad_norm": 1.098029662170612,
      "learning_rate": 0.0001709743270082023,
      "loss": 1.0676,
      "step": 3765
    },
    {
      "epoch": 0.27146255316081597,
      "grad_norm": 0.9377146371718444,
      "learning_rate": 0.00017095787792998613,
      "loss": 1.025,
      "step": 3766
    },
    {
      "epoch": 0.2715346356231529,
      "grad_norm": 0.1458809312799363,
      "learning_rate": 0.00017094142498393078,
      "loss": 1.1074,
      "step": 3767
    },
    {
      "epoch": 0.2716067180854898,
      "grad_norm": 0.1652594951554858,
      "learning_rate": 0.00017092496817093316,
      "loss": 1.0692,
      "step": 3768
    },
    {
      "epoch": 0.2716788005478267,
      "grad_norm": 0.11796492629900103,
      "learning_rate": 0.00017090850749189022,
      "loss": 1.0676,
      "step": 3769
    },
    {
      "epoch": 0.27175088301016365,
      "grad_norm": 0.13826612724934195,
      "learning_rate": 0.00017089204294769926,
      "loss": 1.0548,
      "step": 3770
    },
    {
      "epoch": 0.27182296547250057,
      "grad_norm": 0.12472156118707216,
      "learning_rate": 0.00017087557453925774,
      "loss": 1.1606,
      "step": 3771
    },
    {
      "epoch": 0.27189504793483743,
      "grad_norm": 0.1873933181743621,
      "learning_rate": 0.00017085910226746336,
      "loss": 0.9975,
      "step": 3772
    },
    {
      "epoch": 0.27196713039717435,
      "grad_norm": 0.15258895667663808,
      "learning_rate": 0.00017084262613321398,
      "loss": 1.0439,
      "step": 3773
    },
    {
      "epoch": 0.27203921285951127,
      "grad_norm": 0.11860314384060058,
      "learning_rate": 0.00017082614613740768,
      "loss": 1.1277,
      "step": 3774
    },
    {
      "epoch": 0.2721112953218482,
      "grad_norm": 0.16514369352120717,
      "learning_rate": 0.0001708096622809428,
      "loss": 1.1295,
      "step": 3775
    },
    {
      "epoch": 0.2721833777841851,
      "grad_norm": 0.20340286376969677,
      "learning_rate": 0.00017079317456471785,
      "loss": 1.0966,
      "step": 3776
    },
    {
      "epoch": 0.27225546024652203,
      "grad_norm": 0.13873512633024523,
      "learning_rate": 0.00017077668298963155,
      "loss": 1.0547,
      "step": 3777
    },
    {
      "epoch": 0.27232754270885895,
      "grad_norm": 0.1195340254359444,
      "learning_rate": 0.00017076018755658284,
      "loss": 0.9606,
      "step": 3778
    },
    {
      "epoch": 0.27239962517119587,
      "grad_norm": 0.19048817655935227,
      "learning_rate": 0.00017074368826647084,
      "loss": 1.0608,
      "step": 3779
    },
    {
      "epoch": 0.2724717076335328,
      "grad_norm": 0.1546017364041658,
      "learning_rate": 0.000170727185120195,
      "loss": 0.9891,
      "step": 3780
    },
    {
      "epoch": 0.27254379009586965,
      "grad_norm": 0.15703545235270042,
      "learning_rate": 0.00017071067811865476,
      "loss": 1.0023,
      "step": 3781
    },
    {
      "epoch": 0.2726158725582066,
      "grad_norm": 0.13802095781454143,
      "learning_rate": 0.00017069416726275,
      "loss": 1.0388,
      "step": 3782
    },
    {
      "epoch": 0.2726879550205435,
      "grad_norm": 0.2399938970775095,
      "learning_rate": 0.00017067765255338068,
      "loss": 1.0535,
      "step": 3783
    },
    {
      "epoch": 0.2727600374828804,
      "grad_norm": 0.14174538987087024,
      "learning_rate": 0.000170661133991447,
      "loss": 1.1285,
      "step": 3784
    },
    {
      "epoch": 0.27283211994521733,
      "grad_norm": 0.16225223875499234,
      "learning_rate": 0.00017064461157784933,
      "loss": 1.1003,
      "step": 3785
    },
    {
      "epoch": 0.27290420240755425,
      "grad_norm": 0.14676228024454463,
      "learning_rate": 0.00017062808531348838,
      "loss": 1.0551,
      "step": 3786
    },
    {
      "epoch": 0.2729762848698912,
      "grad_norm": 0.1415213698444885,
      "learning_rate": 0.00017061155519926485,
      "loss": 1.1057,
      "step": 3787
    },
    {
      "epoch": 0.2730483673322281,
      "grad_norm": 0.13729804785830693,
      "learning_rate": 0.0001705950212360799,
      "loss": 1.0815,
      "step": 3788
    },
    {
      "epoch": 0.27312044979456496,
      "grad_norm": 0.16026075540689894,
      "learning_rate": 0.00017057848342483473,
      "loss": 1.0795,
      "step": 3789
    },
    {
      "epoch": 0.2731925322569019,
      "grad_norm": 0.15435872423628172,
      "learning_rate": 0.0001705619417664308,
      "loss": 1.0681,
      "step": 3790
    },
    {
      "epoch": 0.2732646147192388,
      "grad_norm": 0.18772248035364936,
      "learning_rate": 0.00017054539626176976,
      "loss": 1.0655,
      "step": 3791
    },
    {
      "epoch": 0.2733366971815757,
      "grad_norm": 0.12508539824564863,
      "learning_rate": 0.00017052884691175354,
      "loss": 1.11,
      "step": 3792
    },
    {
      "epoch": 0.27340877964391264,
      "grad_norm": 0.11687345642814993,
      "learning_rate": 0.0001705122937172842,
      "loss": 1.1282,
      "step": 3793
    },
    {
      "epoch": 0.27348086210624956,
      "grad_norm": 0.12661988292747722,
      "learning_rate": 0.00017049573667926398,
      "loss": 1.0031,
      "step": 3794
    },
    {
      "epoch": 0.2735529445685865,
      "grad_norm": 0.12911514156250778,
      "learning_rate": 0.00017047917579859548,
      "loss": 1.1289,
      "step": 3795
    },
    {
      "epoch": 0.2736250270309234,
      "grad_norm": 0.29476068158844815,
      "learning_rate": 0.00017046261107618132,
      "loss": 1.1592,
      "step": 3796
    },
    {
      "epoch": 0.2736971094932603,
      "grad_norm": 0.16992270366380324,
      "learning_rate": 0.00017044604251292454,
      "loss": 1.1198,
      "step": 3797
    },
    {
      "epoch": 0.2737691919555972,
      "grad_norm": 0.3748598368706846,
      "learning_rate": 0.00017042947010972818,
      "loss": 1.1058,
      "step": 3798
    },
    {
      "epoch": 0.2738412744179341,
      "grad_norm": 0.13164947443542377,
      "learning_rate": 0.00017041289386749564,
      "loss": 1.0249,
      "step": 3799
    },
    {
      "epoch": 0.273913356880271,
      "grad_norm": 0.13505298721440046,
      "learning_rate": 0.00017039631378713046,
      "loss": 0.9211,
      "step": 3800
    },
    {
      "epoch": 0.27398543934260794,
      "grad_norm": 0.1145985531853881,
      "learning_rate": 0.00017037972986953635,
      "loss": 1.065,
      "step": 3801
    },
    {
      "epoch": 0.27405752180494486,
      "grad_norm": 0.11539786775407444,
      "learning_rate": 0.00017036314211561738,
      "loss": 1.018,
      "step": 3802
    },
    {
      "epoch": 0.2741296042672818,
      "grad_norm": 0.15355633554258508,
      "learning_rate": 0.00017034655052627765,
      "loss": 1.0871,
      "step": 3803
    },
    {
      "epoch": 0.2742016867296187,
      "grad_norm": 0.16672584422673528,
      "learning_rate": 0.0001703299551024216,
      "loss": 1.1412,
      "step": 3804
    },
    {
      "epoch": 0.2742737691919556,
      "grad_norm": 0.12455913386285605,
      "learning_rate": 0.00017031335584495377,
      "loss": 1.1558,
      "step": 3805
    },
    {
      "epoch": 0.2743458516542925,
      "grad_norm": 0.1321801946736055,
      "learning_rate": 0.00017029675275477904,
      "loss": 1.1241,
      "step": 3806
    },
    {
      "epoch": 0.2744179341166294,
      "grad_norm": 0.22215316780442879,
      "learning_rate": 0.00017028014583280235,
      "loss": 1.0568,
      "step": 3807
    },
    {
      "epoch": 0.2744900165789663,
      "grad_norm": 0.15614581433927155,
      "learning_rate": 0.000170263535079929,
      "loss": 1.0685,
      "step": 3808
    },
    {
      "epoch": 0.27456209904130324,
      "grad_norm": 0.13567314139041461,
      "learning_rate": 0.00017024692049706438,
      "loss": 1.0672,
      "step": 3809
    },
    {
      "epoch": 0.27463418150364016,
      "grad_norm": 0.16728747278482767,
      "learning_rate": 0.0001702303020851142,
      "loss": 1.0021,
      "step": 3810
    },
    {
      "epoch": 0.2747062639659771,
      "grad_norm": 0.11251247956807009,
      "learning_rate": 0.0001702136798449842,
      "loss": 1.0697,
      "step": 3811
    },
    {
      "epoch": 0.274778346428314,
      "grad_norm": 0.1423393160081117,
      "learning_rate": 0.0001701970537775805,
      "loss": 0.9595,
      "step": 3812
    },
    {
      "epoch": 0.2748504288906509,
      "grad_norm": 0.12175776566943192,
      "learning_rate": 0.00017018042388380937,
      "loss": 1.0393,
      "step": 3813
    },
    {
      "epoch": 0.27492251135298784,
      "grad_norm": 0.20320426109850293,
      "learning_rate": 0.0001701637901645773,
      "loss": 1.02,
      "step": 3814
    },
    {
      "epoch": 0.2749945938153247,
      "grad_norm": 0.20788591666703932,
      "learning_rate": 0.00017014715262079095,
      "loss": 1.0807,
      "step": 3815
    },
    {
      "epoch": 0.2750666762776616,
      "grad_norm": 0.22123935853899307,
      "learning_rate": 0.00017013051125335725,
      "loss": 1.0053,
      "step": 3816
    },
    {
      "epoch": 0.27513875873999855,
      "grad_norm": 0.239171130190493,
      "learning_rate": 0.00017011386606318328,
      "loss": 1.0431,
      "step": 3817
    },
    {
      "epoch": 0.27521084120233547,
      "grad_norm": 0.11394162855112884,
      "learning_rate": 0.00017009721705117632,
      "loss": 1.1002,
      "step": 3818
    },
    {
      "epoch": 0.2752829236646724,
      "grad_norm": 0.13996451630969453,
      "learning_rate": 0.00017008056421824396,
      "loss": 1.0923,
      "step": 3819
    },
    {
      "epoch": 0.2753550061270093,
      "grad_norm": 0.19445312820212463,
      "learning_rate": 0.00017006390756529383,
      "loss": 1.1866,
      "step": 3820
    },
    {
      "epoch": 0.2754270885893462,
      "grad_norm": 0.12978425593754703,
      "learning_rate": 0.000170047247093234,
      "loss": 1.0434,
      "step": 3821
    },
    {
      "epoch": 0.27549917105168314,
      "grad_norm": 0.8414825439043685,
      "learning_rate": 0.0001700305828029725,
      "loss": 1.0028,
      "step": 3822
    },
    {
      "epoch": 0.27557125351402006,
      "grad_norm": 0.1423137931916084,
      "learning_rate": 0.00017001391469541777,
      "loss": 1.0427,
      "step": 3823
    },
    {
      "epoch": 0.27564333597635693,
      "grad_norm": 0.1367976159589645,
      "learning_rate": 0.0001699972427714783,
      "loss": 1.0825,
      "step": 3824
    },
    {
      "epoch": 0.27571541843869385,
      "grad_norm": 0.15239294603878725,
      "learning_rate": 0.00016998056703206287,
      "loss": 0.9895,
      "step": 3825
    },
    {
      "epoch": 0.27578750090103077,
      "grad_norm": 0.22335415853087642,
      "learning_rate": 0.0001699638874780805,
      "loss": 1.0737,
      "step": 3826
    },
    {
      "epoch": 0.2758595833633677,
      "grad_norm": 0.24854202217110968,
      "learning_rate": 0.00016994720411044032,
      "loss": 1.0918,
      "step": 3827
    },
    {
      "epoch": 0.2759316658257046,
      "grad_norm": 0.13846950628210358,
      "learning_rate": 0.00016993051693005176,
      "loss": 1.1391,
      "step": 3828
    },
    {
      "epoch": 0.27600374828804153,
      "grad_norm": 0.14256262345663687,
      "learning_rate": 0.00016991382593782442,
      "loss": 1.0667,
      "step": 3829
    },
    {
      "epoch": 0.27607583075037845,
      "grad_norm": 0.12848368434479487,
      "learning_rate": 0.0001698971311346681,
      "loss": 1.0427,
      "step": 3830
    },
    {
      "epoch": 0.27614791321271537,
      "grad_norm": 0.13498215051534282,
      "learning_rate": 0.0001698804325214928,
      "loss": 1.0355,
      "step": 3831
    },
    {
      "epoch": 0.27621999567505223,
      "grad_norm": 0.1817044621858368,
      "learning_rate": 0.0001698637300992088,
      "loss": 1.0501,
      "step": 3832
    },
    {
      "epoch": 0.27629207813738915,
      "grad_norm": 0.1494989678629778,
      "learning_rate": 0.00016984702386872643,
      "loss": 1.0385,
      "step": 3833
    },
    {
      "epoch": 0.27636416059972607,
      "grad_norm": 0.21352488263942807,
      "learning_rate": 0.00016983031383095644,
      "loss": 1.0403,
      "step": 3834
    },
    {
      "epoch": 0.276436243062063,
      "grad_norm": 0.12959769392634304,
      "learning_rate": 0.00016981359998680958,
      "loss": 1.045,
      "step": 3835
    },
    {
      "epoch": 0.2765083255243999,
      "grad_norm": 0.16798129210670945,
      "learning_rate": 0.00016979688233719698,
      "loss": 1.0161,
      "step": 3836
    },
    {
      "epoch": 0.27658040798673683,
      "grad_norm": 0.1468330537152582,
      "learning_rate": 0.00016978016088302987,
      "loss": 1.1445,
      "step": 3837
    },
    {
      "epoch": 0.27665249044907375,
      "grad_norm": 0.1484039979585629,
      "learning_rate": 0.00016976343562521976,
      "loss": 1.0202,
      "step": 3838
    },
    {
      "epoch": 0.27672457291141067,
      "grad_norm": 0.14426681064551006,
      "learning_rate": 0.00016974670656467824,
      "loss": 1.0356,
      "step": 3839
    },
    {
      "epoch": 0.2767966553737476,
      "grad_norm": 0.24737299481117175,
      "learning_rate": 0.00016972997370231724,
      "loss": 1.1211,
      "step": 3840
    },
    {
      "epoch": 0.27686873783608446,
      "grad_norm": 0.15387663894644424,
      "learning_rate": 0.00016971323703904884,
      "loss": 1.0996,
      "step": 3841
    },
    {
      "epoch": 0.2769408202984214,
      "grad_norm": 0.16480637720723618,
      "learning_rate": 0.00016969649657578537,
      "loss": 1.0679,
      "step": 3842
    },
    {
      "epoch": 0.2770129027607583,
      "grad_norm": 0.6716394895204048,
      "learning_rate": 0.00016967975231343933,
      "loss": 1.0374,
      "step": 3843
    },
    {
      "epoch": 0.2770849852230952,
      "grad_norm": 0.2788516647280454,
      "learning_rate": 0.0001696630042529234,
      "loss": 1.1194,
      "step": 3844
    },
    {
      "epoch": 0.27715706768543213,
      "grad_norm": 0.1539916479830738,
      "learning_rate": 0.0001696462523951505,
      "loss": 1.111,
      "step": 3845
    },
    {
      "epoch": 0.27722915014776905,
      "grad_norm": 0.12295880678639833,
      "learning_rate": 0.00016962949674103378,
      "loss": 1.1662,
      "step": 3846
    },
    {
      "epoch": 0.277301232610106,
      "grad_norm": 0.12134886811876293,
      "learning_rate": 0.0001696127372914866,
      "loss": 1.077,
      "step": 3847
    },
    {
      "epoch": 0.2773733150724429,
      "grad_norm": 0.19642187279718287,
      "learning_rate": 0.0001695959740474224,
      "loss": 1.1438,
      "step": 3848
    },
    {
      "epoch": 0.2774453975347798,
      "grad_norm": 0.1434188289570967,
      "learning_rate": 0.00016957920700975504,
      "loss": 1.057,
      "step": 3849
    },
    {
      "epoch": 0.2775174799971167,
      "grad_norm": 0.15781774469734047,
      "learning_rate": 0.00016956243617939843,
      "loss": 1.0457,
      "step": 3850
    },
    {
      "epoch": 0.2775895624594536,
      "grad_norm": 0.14490062368924414,
      "learning_rate": 0.00016954566155726668,
      "loss": 1.0141,
      "step": 3851
    },
    {
      "epoch": 0.2776616449217905,
      "grad_norm": 0.11383658339947515,
      "learning_rate": 0.00016952888314427422,
      "loss": 1.0747,
      "step": 3852
    },
    {
      "epoch": 0.27773372738412744,
      "grad_norm": 0.3791710821771313,
      "learning_rate": 0.0001695121009413356,
      "loss": 1.0428,
      "step": 3853
    },
    {
      "epoch": 0.27780580984646436,
      "grad_norm": 0.10747740082861953,
      "learning_rate": 0.0001694953149493656,
      "loss": 1.0626,
      "step": 3854
    },
    {
      "epoch": 0.2778778923088013,
      "grad_norm": 0.151486300531477,
      "learning_rate": 0.00016947852516927925,
      "loss": 1.1431,
      "step": 3855
    },
    {
      "epoch": 0.2779499747711382,
      "grad_norm": 0.12215482923131118,
      "learning_rate": 0.0001694617316019917,
      "loss": 1.0903,
      "step": 3856
    },
    {
      "epoch": 0.2780220572334751,
      "grad_norm": 0.16284471888236787,
      "learning_rate": 0.00016944493424841833,
      "loss": 0.962,
      "step": 3857
    },
    {
      "epoch": 0.278094139695812,
      "grad_norm": 0.13951774325106236,
      "learning_rate": 0.00016942813310947478,
      "loss": 1.0579,
      "step": 3858
    },
    {
      "epoch": 0.2781662221581489,
      "grad_norm": 0.20044960579510723,
      "learning_rate": 0.00016941132818607685,
      "loss": 0.9973,
      "step": 3859
    },
    {
      "epoch": 0.2782383046204858,
      "grad_norm": 0.270173381832828,
      "learning_rate": 0.00016939451947914057,
      "loss": 1.1197,
      "step": 3860
    },
    {
      "epoch": 0.27831038708282274,
      "grad_norm": 0.11267302652482947,
      "learning_rate": 0.00016937770698958215,
      "loss": 1.0816,
      "step": 3861
    },
    {
      "epoch": 0.27838246954515966,
      "grad_norm": 0.1396574497440693,
      "learning_rate": 0.00016936089071831805,
      "loss": 1.0882,
      "step": 3862
    },
    {
      "epoch": 0.2784545520074966,
      "grad_norm": 0.11207451714983854,
      "learning_rate": 0.00016934407066626484,
      "loss": 1.0716,
      "step": 3863
    },
    {
      "epoch": 0.2785266344698335,
      "grad_norm": 0.21861236033849202,
      "learning_rate": 0.00016932724683433944,
      "loss": 1.0442,
      "step": 3864
    },
    {
      "epoch": 0.2785987169321704,
      "grad_norm": 0.12268501983532543,
      "learning_rate": 0.00016931041922345886,
      "loss": 1.0713,
      "step": 3865
    },
    {
      "epoch": 0.27867079939450734,
      "grad_norm": 1.7032144868219168,
      "learning_rate": 0.0001692935878345404,
      "loss": 1.1348,
      "step": 3866
    },
    {
      "epoch": 0.2787428818568442,
      "grad_norm": 0.12827262610058537,
      "learning_rate": 0.00016927675266850141,
      "loss": 1.0189,
      "step": 3867
    },
    {
      "epoch": 0.2788149643191811,
      "grad_norm": 0.12474623248871812,
      "learning_rate": 0.0001692599137262597,
      "loss": 1.136,
      "step": 3868
    },
    {
      "epoch": 0.27888704678151804,
      "grad_norm": 0.11171792184511137,
      "learning_rate": 0.00016924307100873307,
      "loss": 1.0446,
      "step": 3869
    },
    {
      "epoch": 0.27895912924385496,
      "grad_norm": 0.3587102425934392,
      "learning_rate": 0.00016922622451683958,
      "loss": 1.1766,
      "step": 3870
    },
    {
      "epoch": 0.2790312117061919,
      "grad_norm": 0.12554698347386656,
      "learning_rate": 0.00016920937425149752,
      "loss": 1.1053,
      "step": 3871
    },
    {
      "epoch": 0.2791032941685288,
      "grad_norm": 0.9513594391975031,
      "learning_rate": 0.00016919252021362543,
      "loss": 1.0677,
      "step": 3872
    },
    {
      "epoch": 0.2791753766308657,
      "grad_norm": 0.13331258061078405,
      "learning_rate": 0.000169175662404142,
      "loss": 1.0636,
      "step": 3873
    },
    {
      "epoch": 0.27924745909320264,
      "grad_norm": 0.13731190513423636,
      "learning_rate": 0.00016915880082396604,
      "loss": 1.0688,
      "step": 3874
    },
    {
      "epoch": 0.27931954155553956,
      "grad_norm": 0.2569505101947862,
      "learning_rate": 0.00016914193547401676,
      "loss": 1.1223,
      "step": 3875
    },
    {
      "epoch": 0.2793916240178764,
      "grad_norm": 0.13861932965611407,
      "learning_rate": 0.00016912506635521343,
      "loss": 1.1231,
      "step": 3876
    },
    {
      "epoch": 0.27946370648021335,
      "grad_norm": 0.15322384572567715,
      "learning_rate": 0.00016910819346847554,
      "loss": 1.0323,
      "step": 3877
    },
    {
      "epoch": 0.27953578894255027,
      "grad_norm": 0.312062339776104,
      "learning_rate": 0.00016909131681472293,
      "loss": 0.9765,
      "step": 3878
    },
    {
      "epoch": 0.2796078714048872,
      "grad_norm": 0.2955101959846469,
      "learning_rate": 0.00016907443639487536,
      "loss": 1.082,
      "step": 3879
    },
    {
      "epoch": 0.2796799538672241,
      "grad_norm": 0.5218673385651585,
      "learning_rate": 0.0001690575522098531,
      "loss": 1.0821,
      "step": 3880
    },
    {
      "epoch": 0.279752036329561,
      "grad_norm": 1.6145340222336304,
      "learning_rate": 0.00016904066426057644,
      "loss": 1.0389,
      "step": 3881
    },
    {
      "epoch": 0.27982411879189795,
      "grad_norm": 0.23539811190782195,
      "learning_rate": 0.0001690237725479659,
      "loss": 0.9975,
      "step": 3882
    },
    {
      "epoch": 0.27989620125423487,
      "grad_norm": 0.43249243675623256,
      "learning_rate": 0.00016900687707294224,
      "loss": 1.1525,
      "step": 3883
    },
    {
      "epoch": 0.27996828371657173,
      "grad_norm": 0.16395007266854694,
      "learning_rate": 0.00016898997783642642,
      "loss": 1.0984,
      "step": 3884
    },
    {
      "epoch": 0.28004036617890865,
      "grad_norm": 0.2009607822396417,
      "learning_rate": 0.00016897307483933966,
      "loss": 1.0906,
      "step": 3885
    },
    {
      "epoch": 0.28011244864124557,
      "grad_norm": 0.273300633438528,
      "learning_rate": 0.00016895616808260323,
      "loss": 1.0869,
      "step": 3886
    },
    {
      "epoch": 0.2801845311035825,
      "grad_norm": 0.13582476406199495,
      "learning_rate": 0.00016893925756713874,
      "loss": 1.0784,
      "step": 3887
    },
    {
      "epoch": 0.2802566135659194,
      "grad_norm": 0.5384861129903664,
      "learning_rate": 0.00016892234329386798,
      "loss": 1.1355,
      "step": 3888
    },
    {
      "epoch": 0.28032869602825633,
      "grad_norm": 0.21522102027315673,
      "learning_rate": 0.00016890542526371293,
      "loss": 1.09,
      "step": 3889
    },
    {
      "epoch": 0.28040077849059325,
      "grad_norm": 0.15774454525086845,
      "learning_rate": 0.0001688885034775957,
      "loss": 1.1026,
      "step": 3890
    },
    {
      "epoch": 0.28047286095293017,
      "grad_norm": 0.2027703522589538,
      "learning_rate": 0.00016887157793643877,
      "loss": 1.1142,
      "step": 3891
    },
    {
      "epoch": 0.2805449434152671,
      "grad_norm": 0.24505115514638456,
      "learning_rate": 0.00016885464864116473,
      "loss": 1.0293,
      "step": 3892
    },
    {
      "epoch": 0.28061702587760395,
      "grad_norm": 2.0812448347057724,
      "learning_rate": 0.0001688377155926963,
      "loss": 1.188,
      "step": 3893
    },
    {
      "epoch": 0.2806891083399409,
      "grad_norm": 0.14053390512325703,
      "learning_rate": 0.00016882077879195654,
      "loss": 1.0673,
      "step": 3894
    },
    {
      "epoch": 0.2807611908022778,
      "grad_norm": 0.1663882491856698,
      "learning_rate": 0.00016880383823986866,
      "loss": 0.9769,
      "step": 3895
    },
    {
      "epoch": 0.2808332732646147,
      "grad_norm": 0.15710591012042552,
      "learning_rate": 0.00016878689393735606,
      "loss": 1.0491,
      "step": 3896
    },
    {
      "epoch": 0.28090535572695163,
      "grad_norm": 0.18418098478195905,
      "learning_rate": 0.00016876994588534234,
      "loss": 1.1015,
      "step": 3897
    },
    {
      "epoch": 0.28097743818928855,
      "grad_norm": 0.2456390369327744,
      "learning_rate": 0.00016875299408475132,
      "loss": 1.0383,
      "step": 3898
    },
    {
      "epoch": 0.2810495206516255,
      "grad_norm": 0.825287331762268,
      "learning_rate": 0.0001687360385365071,
      "loss": 1.1018,
      "step": 3899
    },
    {
      "epoch": 0.2811216031139624,
      "grad_norm": 0.3080197949792732,
      "learning_rate": 0.0001687190792415338,
      "loss": 1.1442,
      "step": 3900
    },
    {
      "epoch": 0.2811936855762993,
      "grad_norm": 0.1585846572037229,
      "learning_rate": 0.0001687021162007559,
      "loss": 1.0669,
      "step": 3901
    },
    {
      "epoch": 0.2812657680386362,
      "grad_norm": 0.31712054992729716,
      "learning_rate": 0.00016868514941509807,
      "loss": 1.0685,
      "step": 3902
    },
    {
      "epoch": 0.2813378505009731,
      "grad_norm": 0.12378051456570041,
      "learning_rate": 0.00016866817888548514,
      "loss": 1.0408,
      "step": 3903
    },
    {
      "epoch": 0.28140993296331,
      "grad_norm": 0.13328890239257477,
      "learning_rate": 0.00016865120461284207,
      "loss": 1.1071,
      "step": 3904
    },
    {
      "epoch": 0.28148201542564694,
      "grad_norm": 0.48480990456498163,
      "learning_rate": 0.00016863422659809424,
      "loss": 1.0873,
      "step": 3905
    },
    {
      "epoch": 0.28155409788798386,
      "grad_norm": 0.11675406221799621,
      "learning_rate": 0.00016861724484216702,
      "loss": 1.0526,
      "step": 3906
    },
    {
      "epoch": 0.2816261803503208,
      "grad_norm": 0.12888820713135632,
      "learning_rate": 0.0001686002593459861,
      "loss": 1.0779,
      "step": 3907
    },
    {
      "epoch": 0.2816982628126577,
      "grad_norm": 0.3640493225100968,
      "learning_rate": 0.0001685832701104773,
      "loss": 0.9983,
      "step": 3908
    },
    {
      "epoch": 0.2817703452749946,
      "grad_norm": 0.1774853194180583,
      "learning_rate": 0.00016856627713656672,
      "loss": 1.135,
      "step": 3909
    },
    {
      "epoch": 0.2818424277373315,
      "grad_norm": 0.18376185631943012,
      "learning_rate": 0.00016854928042518062,
      "loss": 1.1256,
      "step": 3910
    },
    {
      "epoch": 0.2819145101996684,
      "grad_norm": 0.12600752040660895,
      "learning_rate": 0.00016853227997724554,
      "loss": 1.0636,
      "step": 3911
    },
    {
      "epoch": 0.2819865926620053,
      "grad_norm": 0.1367123824091876,
      "learning_rate": 0.000168515275793688,
      "loss": 1.0989,
      "step": 3912
    },
    {
      "epoch": 0.28205867512434224,
      "grad_norm": 0.1714260573790089,
      "learning_rate": 0.00016849826787543502,
      "loss": 1.0138,
      "step": 3913
    },
    {
      "epoch": 0.28213075758667916,
      "grad_norm": 0.36343159309505446,
      "learning_rate": 0.00016848125622341366,
      "loss": 1.071,
      "step": 3914
    },
    {
      "epoch": 0.2822028400490161,
      "grad_norm": 0.738508843436725,
      "learning_rate": 0.00016846424083855117,
      "loss": 1.1167,
      "step": 3915
    },
    {
      "epoch": 0.282274922511353,
      "grad_norm": 0.847724369161873,
      "learning_rate": 0.00016844722172177506,
      "loss": 1.0475,
      "step": 3916
    },
    {
      "epoch": 0.2823470049736899,
      "grad_norm": 0.1595397953642581,
      "learning_rate": 0.000168430198874013,
      "loss": 1.1161,
      "step": 3917
    },
    {
      "epoch": 0.28241908743602684,
      "grad_norm": 0.24530502595728512,
      "learning_rate": 0.00016841317229619293,
      "loss": 1.1382,
      "step": 3918
    },
    {
      "epoch": 0.2824911698983637,
      "grad_norm": 0.13142616919103514,
      "learning_rate": 0.0001683961419892429,
      "loss": 1.0702,
      "step": 3919
    },
    {
      "epoch": 0.2825632523607006,
      "grad_norm": 0.12076500339555422,
      "learning_rate": 0.00016837910795409127,
      "loss": 1.1361,
      "step": 3920
    },
    {
      "epoch": 0.28263533482303754,
      "grad_norm": 0.4510969925709476,
      "learning_rate": 0.00016836207019166647,
      "loss": 1.0992,
      "step": 3921
    },
    {
      "epoch": 0.28270741728537446,
      "grad_norm": 0.2190282219473502,
      "learning_rate": 0.00016834502870289732,
      "loss": 1.0728,
      "step": 3922
    },
    {
      "epoch": 0.2827794997477114,
      "grad_norm": 0.22028678681690425,
      "learning_rate": 0.00016832798348871269,
      "loss": 1.0185,
      "step": 3923
    },
    {
      "epoch": 0.2828515822100483,
      "grad_norm": 0.1267302813483403,
      "learning_rate": 0.00016831093455004162,
      "loss": 1.0742,
      "step": 3924
    },
    {
      "epoch": 0.2829236646723852,
      "grad_norm": 0.18349006211673252,
      "learning_rate": 0.00016829388188781353,
      "loss": 1.1361,
      "step": 3925
    },
    {
      "epoch": 0.28299574713472214,
      "grad_norm": 0.18484114951755945,
      "learning_rate": 0.00016827682550295785,
      "loss": 1.0332,
      "step": 3926
    },
    {
      "epoch": 0.28306782959705906,
      "grad_norm": 0.16188164532310387,
      "learning_rate": 0.0001682597653964044,
      "loss": 1.1022,
      "step": 3927
    },
    {
      "epoch": 0.2831399120593959,
      "grad_norm": 0.11812021212378952,
      "learning_rate": 0.00016824270156908305,
      "loss": 0.9869,
      "step": 3928
    },
    {
      "epoch": 0.28321199452173285,
      "grad_norm": 0.11558820928182874,
      "learning_rate": 0.00016822563402192397,
      "loss": 1.0019,
      "step": 3929
    },
    {
      "epoch": 0.28328407698406977,
      "grad_norm": 0.6141863214435059,
      "learning_rate": 0.00016820856275585744,
      "loss": 1.0813,
      "step": 3930
    },
    {
      "epoch": 0.2833561594464067,
      "grad_norm": 0.32014834410756104,
      "learning_rate": 0.00016819148777181406,
      "loss": 1.0347,
      "step": 3931
    },
    {
      "epoch": 0.2834282419087436,
      "grad_norm": 0.1669316997914028,
      "learning_rate": 0.0001681744090707245,
      "loss": 0.9919,
      "step": 3932
    },
    {
      "epoch": 0.2835003243710805,
      "grad_norm": 0.1492387163268774,
      "learning_rate": 0.00016815732665351977,
      "loss": 1.1372,
      "step": 3933
    },
    {
      "epoch": 0.28357240683341745,
      "grad_norm": 0.30021082279841876,
      "learning_rate": 0.000168140240521131,
      "loss": 1.2202,
      "step": 3934
    },
    {
      "epoch": 0.28364448929575437,
      "grad_norm": 0.24000688149843838,
      "learning_rate": 0.0001681231506744895,
      "loss": 1.1377,
      "step": 3935
    },
    {
      "epoch": 0.28371657175809123,
      "grad_norm": 0.19347434971625083,
      "learning_rate": 0.00016810605711452685,
      "loss": 1.1088,
      "step": 3936
    },
    {
      "epoch": 0.28378865422042815,
      "grad_norm": 0.12167338952769356,
      "learning_rate": 0.00016808895984217476,
      "loss": 1.0892,
      "step": 3937
    },
    {
      "epoch": 0.28386073668276507,
      "grad_norm": 0.19583413245918924,
      "learning_rate": 0.00016807185885836526,
      "loss": 1.0903,
      "step": 3938
    },
    {
      "epoch": 0.283932819145102,
      "grad_norm": 0.12425472798942917,
      "learning_rate": 0.00016805475416403046,
      "loss": 1.049,
      "step": 3939
    },
    {
      "epoch": 0.2840049016074389,
      "grad_norm": 0.2575527544437409,
      "learning_rate": 0.00016803764576010268,
      "loss": 1.1061,
      "step": 3940
    },
    {
      "epoch": 0.28407698406977583,
      "grad_norm": 0.21316982378942845,
      "learning_rate": 0.00016802053364751456,
      "loss": 1.0866,
      "step": 3941
    },
    {
      "epoch": 0.28414906653211275,
      "grad_norm": 0.11841122135469419,
      "learning_rate": 0.00016800341782719882,
      "loss": 1.0777,
      "step": 3942
    },
    {
      "epoch": 0.28422114899444967,
      "grad_norm": 0.20660256220721557,
      "learning_rate": 0.00016798629830008842,
      "loss": 0.9912,
      "step": 3943
    },
    {
      "epoch": 0.2842932314567866,
      "grad_norm": 0.19002877434907878,
      "learning_rate": 0.00016796917506711656,
      "loss": 1.0283,
      "step": 3944
    },
    {
      "epoch": 0.28436531391912345,
      "grad_norm": 0.1532480266051294,
      "learning_rate": 0.0001679520481292166,
      "loss": 1.1035,
      "step": 3945
    },
    {
      "epoch": 0.2844373963814604,
      "grad_norm": 0.23268031059748795,
      "learning_rate": 0.00016793491748732207,
      "loss": 1.0462,
      "step": 3946
    },
    {
      "epoch": 0.2845094788437973,
      "grad_norm": 0.1607010594489326,
      "learning_rate": 0.0001679177831423668,
      "loss": 1.0951,
      "step": 3947
    },
    {
      "epoch": 0.2845815613061342,
      "grad_norm": 0.4846941385220313,
      "learning_rate": 0.0001679006450952847,
      "loss": 1.0905,
      "step": 3948
    },
    {
      "epoch": 0.28465364376847113,
      "grad_norm": 0.1305380310299388,
      "learning_rate": 0.00016788350334701,
      "loss": 1.0841,
      "step": 3949
    },
    {
      "epoch": 0.28472572623080805,
      "grad_norm": 0.23697735688794577,
      "learning_rate": 0.0001678663578984771,
      "loss": 1.0708,
      "step": 3950
    },
    {
      "epoch": 0.28479780869314497,
      "grad_norm": 0.107992137449837,
      "learning_rate": 0.00016784920875062052,
      "loss": 0.9804,
      "step": 3951
    },
    {
      "epoch": 0.2848698911554819,
      "grad_norm": 0.15125619050513334,
      "learning_rate": 0.00016783205590437504,
      "loss": 1.1006,
      "step": 3952
    },
    {
      "epoch": 0.2849419736178188,
      "grad_norm": 0.12208682679708217,
      "learning_rate": 0.0001678148993606757,
      "loss": 1.0115,
      "step": 3953
    },
    {
      "epoch": 0.2850140560801557,
      "grad_norm": 0.187250283189242,
      "learning_rate": 0.00016779773912045767,
      "loss": 1.0242,
      "step": 3954
    },
    {
      "epoch": 0.2850861385424926,
      "grad_norm": 0.15361437222182703,
      "learning_rate": 0.0001677805751846563,
      "loss": 0.9756,
      "step": 3955
    },
    {
      "epoch": 0.2851582210048295,
      "grad_norm": 0.22507233729958362,
      "learning_rate": 0.0001677634075542072,
      "loss": 1.0993,
      "step": 3956
    },
    {
      "epoch": 0.28523030346716644,
      "grad_norm": 0.11412270458110406,
      "learning_rate": 0.00016774623623004616,
      "loss": 0.9861,
      "step": 3957
    },
    {
      "epoch": 0.28530238592950335,
      "grad_norm": 0.13354540268673085,
      "learning_rate": 0.00016772906121310919,
      "loss": 1.0012,
      "step": 3958
    },
    {
      "epoch": 0.2853744683918403,
      "grad_norm": 0.147506550899467,
      "learning_rate": 0.00016771188250433246,
      "loss": 1.0828,
      "step": 3959
    },
    {
      "epoch": 0.2854465508541772,
      "grad_norm": 1.1023782119649328,
      "learning_rate": 0.00016769470010465234,
      "loss": 1.1744,
      "step": 3960
    },
    {
      "epoch": 0.2855186333165141,
      "grad_norm": 0.23507221238194356,
      "learning_rate": 0.00016767751401500545,
      "loss": 1.0779,
      "step": 3961
    },
    {
      "epoch": 0.285590715778851,
      "grad_norm": 0.17175354338655444,
      "learning_rate": 0.0001676603242363286,
      "loss": 1.0481,
      "step": 3962
    },
    {
      "epoch": 0.2856627982411879,
      "grad_norm": 0.16319075910963735,
      "learning_rate": 0.00016764313076955876,
      "loss": 1.0185,
      "step": 3963
    },
    {
      "epoch": 0.2857348807035248,
      "grad_norm": 0.1324762208641695,
      "learning_rate": 0.00016762593361563318,
      "loss": 1.0455,
      "step": 3964
    },
    {
      "epoch": 0.28580696316586174,
      "grad_norm": 0.17681682420888448,
      "learning_rate": 0.00016760873277548916,
      "loss": 1.0666,
      "step": 3965
    },
    {
      "epoch": 0.28587904562819866,
      "grad_norm": 0.31741130119202854,
      "learning_rate": 0.0001675915282500644,
      "loss": 1.0848,
      "step": 3966
    },
    {
      "epoch": 0.2859511280905356,
      "grad_norm": 0.7170053201246389,
      "learning_rate": 0.00016757432004029665,
      "loss": 1.1418,
      "step": 3967
    },
    {
      "epoch": 0.2860232105528725,
      "grad_norm": 0.41093986393116216,
      "learning_rate": 0.00016755710814712394,
      "loss": 1.1157,
      "step": 3968
    },
    {
      "epoch": 0.2860952930152094,
      "grad_norm": 0.1732009718537187,
      "learning_rate": 0.00016753989257148438,
      "loss": 1.0364,
      "step": 3969
    },
    {
      "epoch": 0.28616737547754634,
      "grad_norm": 0.12283406690347176,
      "learning_rate": 0.00016752267331431652,
      "loss": 1.1155,
      "step": 3970
    },
    {
      "epoch": 0.2862394579398832,
      "grad_norm": 0.13830613989672771,
      "learning_rate": 0.00016750545037655886,
      "loss": 1.1268,
      "step": 3971
    },
    {
      "epoch": 0.2863115404022201,
      "grad_norm": 0.3526416364791347,
      "learning_rate": 0.00016748822375915027,
      "loss": 1.0481,
      "step": 3972
    },
    {
      "epoch": 0.28638362286455704,
      "grad_norm": 0.13862268090654656,
      "learning_rate": 0.0001674709934630297,
      "loss": 1.0428,
      "step": 3973
    },
    {
      "epoch": 0.28645570532689396,
      "grad_norm": 1.1257683778874472,
      "learning_rate": 0.00016745375948913632,
      "loss": 1.0822,
      "step": 3974
    },
    {
      "epoch": 0.2865277877892309,
      "grad_norm": 0.3305793301228597,
      "learning_rate": 0.00016743652183840964,
      "loss": 1.1076,
      "step": 3975
    },
    {
      "epoch": 0.2865998702515678,
      "grad_norm": 0.2428319694744295,
      "learning_rate": 0.0001674192805117892,
      "loss": 1.1302,
      "step": 3976
    },
    {
      "epoch": 0.2866719527139047,
      "grad_norm": 0.38707808213614164,
      "learning_rate": 0.00016740203551021484,
      "loss": 1.1279,
      "step": 3977
    },
    {
      "epoch": 0.28674403517624164,
      "grad_norm": 0.15004026124444295,
      "learning_rate": 0.00016738478683462656,
      "loss": 1.0897,
      "step": 3978
    },
    {
      "epoch": 0.28681611763857856,
      "grad_norm": 0.1433962073482034,
      "learning_rate": 0.00016736753448596452,
      "loss": 1.0643,
      "step": 3979
    },
    {
      "epoch": 0.2868882001009154,
      "grad_norm": 0.12168076989086156,
      "learning_rate": 0.00016735027846516917,
      "loss": 1.1448,
      "step": 3980
    },
    {
      "epoch": 0.28696028256325234,
      "grad_norm": 0.19962932762513355,
      "learning_rate": 0.0001673330187731811,
      "loss": 1.1145,
      "step": 3981
    },
    {
      "epoch": 0.28703236502558926,
      "grad_norm": 0.16793794292299088,
      "learning_rate": 0.00016731575541094115,
      "loss": 1.0892,
      "step": 3982
    },
    {
      "epoch": 0.2871044474879262,
      "grad_norm": 0.19434094456275555,
      "learning_rate": 0.0001672984883793903,
      "loss": 1.0808,
      "step": 3983
    },
    {
      "epoch": 0.2871765299502631,
      "grad_norm": 0.13784176689284589,
      "learning_rate": 0.00016728121767946977,
      "loss": 1.0857,
      "step": 3984
    },
    {
      "epoch": 0.2872486124126,
      "grad_norm": 0.13907712326539207,
      "learning_rate": 0.00016726394331212095,
      "loss": 1.0933,
      "step": 3985
    },
    {
      "epoch": 0.28732069487493694,
      "grad_norm": 0.20842591392361418,
      "learning_rate": 0.00016724666527828547,
      "loss": 1.0297,
      "step": 3986
    },
    {
      "epoch": 0.28739277733727386,
      "grad_norm": 1.7388092846261818,
      "learning_rate": 0.00016722938357890512,
      "loss": 1.2195,
      "step": 3987
    },
    {
      "epoch": 0.28746485979961073,
      "grad_norm": 0.3511758121790119,
      "learning_rate": 0.00016721209821492192,
      "loss": 1.0748,
      "step": 3988
    },
    {
      "epoch": 0.28753694226194765,
      "grad_norm": 0.23562492388433262,
      "learning_rate": 0.00016719480918727807,
      "loss": 1.0268,
      "step": 3989
    },
    {
      "epoch": 0.28760902472428457,
      "grad_norm": 0.13114837718316302,
      "learning_rate": 0.00016717751649691597,
      "loss": 1.0323,
      "step": 3990
    },
    {
      "epoch": 0.2876811071866215,
      "grad_norm": 0.2455103833334105,
      "learning_rate": 0.00016716022014477827,
      "loss": 1.0739,
      "step": 3991
    },
    {
      "epoch": 0.2877531896489584,
      "grad_norm": 0.16061709300092814,
      "learning_rate": 0.00016714292013180767,
      "loss": 1.0673,
      "step": 3992
    },
    {
      "epoch": 0.2878252721112953,
      "grad_norm": 0.8302029175283475,
      "learning_rate": 0.0001671256164589473,
      "loss": 1.1143,
      "step": 3993
    },
    {
      "epoch": 0.28789735457363225,
      "grad_norm": 0.1880661760062113,
      "learning_rate": 0.0001671083091271403,
      "loss": 1.0543,
      "step": 3994
    },
    {
      "epoch": 0.28796943703596917,
      "grad_norm": 0.19088756119353012,
      "learning_rate": 0.00016709099813733006,
      "loss": 1.0593,
      "step": 3995
    },
    {
      "epoch": 0.2880415194983061,
      "grad_norm": 0.5505613435315944,
      "learning_rate": 0.00016707368349046022,
      "loss": 1.0467,
      "step": 3996
    },
    {
      "epoch": 0.28811360196064295,
      "grad_norm": 0.3499832816915443,
      "learning_rate": 0.0001670563651874746,
      "loss": 1.1138,
      "step": 3997
    },
    {
      "epoch": 0.28818568442297987,
      "grad_norm": 0.3529672333031102,
      "learning_rate": 0.00016703904322931714,
      "loss": 1.1476,
      "step": 3998
    },
    {
      "epoch": 0.2882577668853168,
      "grad_norm": 0.21223301373484638,
      "learning_rate": 0.00016702171761693208,
      "loss": 1.0452,
      "step": 3999
    },
    {
      "epoch": 0.2883298493476537,
      "grad_norm": 0.285051865850142,
      "learning_rate": 0.00016700438835126383,
      "loss": 1.0456,
      "step": 4000
    },
    {
      "epoch": 0.28840193180999063,
      "grad_norm": 0.521311725992712,
      "learning_rate": 0.00016698705543325698,
      "loss": 1.0763,
      "step": 4001
    },
    {
      "epoch": 0.28847401427232755,
      "grad_norm": 0.15550599128304804,
      "learning_rate": 0.00016696971886385632,
      "loss": 1.0351,
      "step": 4002
    },
    {
      "epoch": 0.28854609673466447,
      "grad_norm": 0.26683261102703043,
      "learning_rate": 0.00016695237864400682,
      "loss": 1.1073,
      "step": 4003
    },
    {
      "epoch": 0.2886181791970014,
      "grad_norm": 0.13628509077152004,
      "learning_rate": 0.00016693503477465375,
      "loss": 1.0792,
      "step": 4004
    },
    {
      "epoch": 0.2886902616593383,
      "grad_norm": 0.12057066405733811,
      "learning_rate": 0.00016691768725674247,
      "loss": 1.1565,
      "step": 4005
    },
    {
      "epoch": 0.2887623441216752,
      "grad_norm": 0.15000476682379213,
      "learning_rate": 0.00016690033609121857,
      "loss": 1.0946,
      "step": 4006
    },
    {
      "epoch": 0.2888344265840121,
      "grad_norm": 0.12214761489037712,
      "learning_rate": 0.0001668829812790278,
      "loss": 1.1247,
      "step": 4007
    },
    {
      "epoch": 0.288906509046349,
      "grad_norm": 0.21495530687364162,
      "learning_rate": 0.00016686562282111625,
      "loss": 1.0948,
      "step": 4008
    },
    {
      "epoch": 0.28897859150868593,
      "grad_norm": 0.33094874022169535,
      "learning_rate": 0.00016684826071843004,
      "loss": 1.1647,
      "step": 4009
    },
    {
      "epoch": 0.28905067397102285,
      "grad_norm": 0.36784224989771264,
      "learning_rate": 0.0001668308949719156,
      "loss": 1.1026,
      "step": 4010
    },
    {
      "epoch": 0.2891227564333598,
      "grad_norm": 0.13004613102958834,
      "learning_rate": 0.00016681352558251946,
      "loss": 1.0911,
      "step": 4011
    },
    {
      "epoch": 0.2891948388956967,
      "grad_norm": 0.370380466740441,
      "learning_rate": 0.00016679615255118846,
      "loss": 1.0876,
      "step": 4012
    },
    {
      "epoch": 0.2892669213580336,
      "grad_norm": 1.8663995096288535,
      "learning_rate": 0.00016677877587886956,
      "loss": 1.0971,
      "step": 4013
    },
    {
      "epoch": 0.2893390038203705,
      "grad_norm": 0.21715552898122598,
      "learning_rate": 0.00016676139556651,
      "loss": 1.0286,
      "step": 4014
    },
    {
      "epoch": 0.2894110862827074,
      "grad_norm": 0.238898232822798,
      "learning_rate": 0.00016674401161505706,
      "loss": 1.1647,
      "step": 4015
    },
    {
      "epoch": 0.2894831687450443,
      "grad_norm": 0.30014200240814753,
      "learning_rate": 0.0001667266240254584,
      "loss": 1.0893,
      "step": 4016
    },
    {
      "epoch": 0.28955525120738124,
      "grad_norm": 0.16018334922001684,
      "learning_rate": 0.0001667092327986618,
      "loss": 1.0396,
      "step": 4017
    },
    {
      "epoch": 0.28962733366971816,
      "grad_norm": 0.2469556361964772,
      "learning_rate": 0.0001666918379356152,
      "loss": 1.041,
      "step": 4018
    },
    {
      "epoch": 0.2896994161320551,
      "grad_norm": 0.16581557561812388,
      "learning_rate": 0.00016667443943726678,
      "loss": 1.0945,
      "step": 4019
    },
    {
      "epoch": 0.289771498594392,
      "grad_norm": 0.3046213981951856,
      "learning_rate": 0.00016665703730456491,
      "loss": 1.0953,
      "step": 4020
    },
    {
      "epoch": 0.2898435810567289,
      "grad_norm": 0.12113579495692062,
      "learning_rate": 0.0001666396315384582,
      "loss": 1.0765,
      "step": 4021
    },
    {
      "epoch": 0.28991566351906584,
      "grad_norm": 0.14162721173474366,
      "learning_rate": 0.0001666222221398954,
      "loss": 1.0718,
      "step": 4022
    },
    {
      "epoch": 0.2899877459814027,
      "grad_norm": 0.135653919803506,
      "learning_rate": 0.00016660480910982546,
      "loss": 0.9936,
      "step": 4023
    },
    {
      "epoch": 0.2900598284437396,
      "grad_norm": 0.14108807942524934,
      "learning_rate": 0.0001665873924491976,
      "loss": 1.0741,
      "step": 4024
    },
    {
      "epoch": 0.29013191090607654,
      "grad_norm": 0.15152644137715907,
      "learning_rate": 0.0001665699721589611,
      "loss": 1.0476,
      "step": 4025
    },
    {
      "epoch": 0.29020399336841346,
      "grad_norm": 0.18978855947312484,
      "learning_rate": 0.00016655254824006556,
      "loss": 1.0792,
      "step": 4026
    },
    {
      "epoch": 0.2902760758307504,
      "grad_norm": 0.13756146563119834,
      "learning_rate": 0.00016653512069346075,
      "loss": 1.0709,
      "step": 4027
    },
    {
      "epoch": 0.2903481582930873,
      "grad_norm": 0.16130331750011861,
      "learning_rate": 0.00016651768952009662,
      "loss": 0.9898,
      "step": 4028
    },
    {
      "epoch": 0.2904202407554242,
      "grad_norm": 0.14524602145585816,
      "learning_rate": 0.00016650025472092334,
      "loss": 1.1237,
      "step": 4029
    },
    {
      "epoch": 0.29049232321776114,
      "grad_norm": 0.2961518561537343,
      "learning_rate": 0.00016648281629689127,
      "loss": 1.1149,
      "step": 4030
    },
    {
      "epoch": 0.29056440568009806,
      "grad_norm": 0.9915978442268314,
      "learning_rate": 0.0001664653742489509,
      "loss": 1.0997,
      "step": 4031
    },
    {
      "epoch": 0.2906364881424349,
      "grad_norm": 0.13118344767439508,
      "learning_rate": 0.00016644792857805306,
      "loss": 1.0813,
      "step": 4032
    },
    {
      "epoch": 0.29070857060477184,
      "grad_norm": 0.1314693306040667,
      "learning_rate": 0.00016643047928514861,
      "loss": 1.0364,
      "step": 4033
    },
    {
      "epoch": 0.29078065306710876,
      "grad_norm": 0.13611906203076946,
      "learning_rate": 0.00016641302637118873,
      "loss": 1.0929,
      "step": 4034
    },
    {
      "epoch": 0.2908527355294457,
      "grad_norm": 0.16594194772938559,
      "learning_rate": 0.00016639556983712481,
      "loss": 1.0924,
      "step": 4035
    },
    {
      "epoch": 0.2909248179917826,
      "grad_norm": 0.12764761715592718,
      "learning_rate": 0.00016637810968390832,
      "loss": 1.0369,
      "step": 4036
    },
    {
      "epoch": 0.2909969004541195,
      "grad_norm": 0.12738395869222222,
      "learning_rate": 0.000166360645912491,
      "loss": 1.0964,
      "step": 4037
    },
    {
      "epoch": 0.29106898291645644,
      "grad_norm": 0.17187908783525674,
      "learning_rate": 0.00016634317852382482,
      "loss": 1.0755,
      "step": 4038
    },
    {
      "epoch": 0.29114106537879336,
      "grad_norm": 0.13793554108830014,
      "learning_rate": 0.00016632570751886187,
      "loss": 1.1387,
      "step": 4039
    },
    {
      "epoch": 0.2912131478411302,
      "grad_norm": 0.12828248403388584,
      "learning_rate": 0.0001663082328985545,
      "loss": 1.0391,
      "step": 4040
    },
    {
      "epoch": 0.29128523030346715,
      "grad_norm": 0.11350779770246215,
      "learning_rate": 0.00016629075466385523,
      "loss": 1.109,
      "step": 4041
    },
    {
      "epoch": 0.29135731276580407,
      "grad_norm": 0.1493512886592976,
      "learning_rate": 0.00016627327281571678,
      "loss": 1.0857,
      "step": 4042
    },
    {
      "epoch": 0.291429395228141,
      "grad_norm": 0.1345428276331405,
      "learning_rate": 0.00016625578735509207,
      "loss": 1.0285,
      "step": 4043
    },
    {
      "epoch": 0.2915014776904779,
      "grad_norm": 0.4625081836321701,
      "learning_rate": 0.0001662382982829342,
      "loss": 1.1085,
      "step": 4044
    },
    {
      "epoch": 0.2915735601528148,
      "grad_norm": 0.5303522851934609,
      "learning_rate": 0.00016622080560019651,
      "loss": 1.118,
      "step": 4045
    },
    {
      "epoch": 0.29164564261515175,
      "grad_norm": 0.3640626684045063,
      "learning_rate": 0.00016620330930783243,
      "loss": 1.0637,
      "step": 4046
    },
    {
      "epoch": 0.29171772507748867,
      "grad_norm": 0.30382917599193904,
      "learning_rate": 0.00016618580940679577,
      "loss": 1.1003,
      "step": 4047
    },
    {
      "epoch": 0.2917898075398256,
      "grad_norm": 0.1691325010668621,
      "learning_rate": 0.0001661683058980404,
      "loss": 1.038,
      "step": 4048
    },
    {
      "epoch": 0.29186189000216245,
      "grad_norm": 0.30193404577381044,
      "learning_rate": 0.00016615079878252033,
      "loss": 1.1557,
      "step": 4049
    },
    {
      "epoch": 0.29193397246449937,
      "grad_norm": 0.11176760519350717,
      "learning_rate": 0.00016613328806118998,
      "loss": 1.0557,
      "step": 4050
    },
    {
      "epoch": 0.2920060549268363,
      "grad_norm": 0.1259541466098484,
      "learning_rate": 0.00016611577373500377,
      "loss": 1.0701,
      "step": 4051
    },
    {
      "epoch": 0.2920781373891732,
      "grad_norm": 0.17296159859587731,
      "learning_rate": 0.00016609825580491641,
      "loss": 1.0443,
      "step": 4052
    },
    {
      "epoch": 0.29215021985151013,
      "grad_norm": 0.26149364359019533,
      "learning_rate": 0.00016608073427188273,
      "loss": 1.0862,
      "step": 4053
    },
    {
      "epoch": 0.29222230231384705,
      "grad_norm": 0.15755113681059962,
      "learning_rate": 0.0001660632091368579,
      "loss": 1.1122,
      "step": 4054
    },
    {
      "epoch": 0.29229438477618397,
      "grad_norm": 1.610847416224277,
      "learning_rate": 0.00016604568040079717,
      "loss": 1.0712,
      "step": 4055
    },
    {
      "epoch": 0.2923664672385209,
      "grad_norm": 0.9493677156824935,
      "learning_rate": 0.00016602814806465597,
      "loss": 0.9958,
      "step": 4056
    },
    {
      "epoch": 0.2924385497008578,
      "grad_norm": 0.16001113449785478,
      "learning_rate": 0.00016601061212939,
      "loss": 1.1613,
      "step": 4057
    },
    {
      "epoch": 0.2925106321631947,
      "grad_norm": 0.1450730072936479,
      "learning_rate": 0.00016599307259595512,
      "loss": 1.0318,
      "step": 4058
    },
    {
      "epoch": 0.2925827146255316,
      "grad_norm": 0.13864473722659304,
      "learning_rate": 0.00016597552946530738,
      "loss": 1.1598,
      "step": 4059
    },
    {
      "epoch": 0.2926547970878685,
      "grad_norm": 0.17128349719107072,
      "learning_rate": 0.00016595798273840305,
      "loss": 1.1196,
      "step": 4060
    },
    {
      "epoch": 0.29272687955020543,
      "grad_norm": 0.24425551607735896,
      "learning_rate": 0.0001659404324161986,
      "loss": 1.0477,
      "step": 4061
    },
    {
      "epoch": 0.29279896201254235,
      "grad_norm": 0.29410326870907866,
      "learning_rate": 0.00016592287849965067,
      "loss": 1.0286,
      "step": 4062
    },
    {
      "epoch": 0.29287104447487927,
      "grad_norm": 0.14906822041143095,
      "learning_rate": 0.0001659053209897161,
      "loss": 1.0846,
      "step": 4063
    },
    {
      "epoch": 0.2929431269372162,
      "grad_norm": 0.2770847959483187,
      "learning_rate": 0.00016588775988735191,
      "loss": 1.1278,
      "step": 4064
    },
    {
      "epoch": 0.2930152093995531,
      "grad_norm": 0.15170726227686768,
      "learning_rate": 0.00016587019519351539,
      "loss": 1.0294,
      "step": 4065
    },
    {
      "epoch": 0.29308729186189,
      "grad_norm": 0.17155271555136964,
      "learning_rate": 0.00016585262690916393,
      "loss": 1.0166,
      "step": 4066
    },
    {
      "epoch": 0.2931593743242269,
      "grad_norm": 1.387694920363889,
      "learning_rate": 0.00016583505503525514,
      "loss": 1.0066,
      "step": 4067
    },
    {
      "epoch": 0.2932314567865638,
      "grad_norm": 0.2490204155041511,
      "learning_rate": 0.00016581747957274693,
      "loss": 1.1011,
      "step": 4068
    },
    {
      "epoch": 0.29330353924890074,
      "grad_norm": 0.1621947749265816,
      "learning_rate": 0.0001657999005225972,
      "loss": 1.0334,
      "step": 4069
    },
    {
      "epoch": 0.29337562171123766,
      "grad_norm": 0.23015328519085593,
      "learning_rate": 0.00016578231788576427,
      "loss": 1.0325,
      "step": 4070
    },
    {
      "epoch": 0.2934477041735746,
      "grad_norm": 3.5161304519490892,
      "learning_rate": 0.00016576473166320644,
      "loss": 1.0628,
      "step": 4071
    },
    {
      "epoch": 0.2935197866359115,
      "grad_norm": 0.13793145203013424,
      "learning_rate": 0.00016574714185588248,
      "loss": 1.096,
      "step": 4072
    },
    {
      "epoch": 0.2935918690982484,
      "grad_norm": 0.14482697525495808,
      "learning_rate": 0.000165729548464751,
      "loss": 1.1052,
      "step": 4073
    },
    {
      "epoch": 0.29366395156058533,
      "grad_norm": 0.13394898651401138,
      "learning_rate": 0.00016571195149077118,
      "loss": 1.123,
      "step": 4074
    },
    {
      "epoch": 0.2937360340229222,
      "grad_norm": 0.13502782782224684,
      "learning_rate": 0.0001656943509349021,
      "loss": 1.1921,
      "step": 4075
    },
    {
      "epoch": 0.2938081164852591,
      "grad_norm": 0.2380619716547345,
      "learning_rate": 0.0001656767467981032,
      "loss": 0.9443,
      "step": 4076
    },
    {
      "epoch": 0.29388019894759604,
      "grad_norm": 0.2558989003054534,
      "learning_rate": 0.000165659139081334,
      "loss": 1.0588,
      "step": 4077
    },
    {
      "epoch": 0.29395228140993296,
      "grad_norm": 0.11594703001626211,
      "learning_rate": 0.00016564152778555434,
      "loss": 1.0315,
      "step": 4078
    },
    {
      "epoch": 0.2940243638722699,
      "grad_norm": 0.6061812638761986,
      "learning_rate": 0.00016562391291172416,
      "loss": 1.0585,
      "step": 4079
    },
    {
      "epoch": 0.2940964463346068,
      "grad_norm": 0.1325254849246701,
      "learning_rate": 0.0001656062944608036,
      "loss": 1.0828,
      "step": 4080
    },
    {
      "epoch": 0.2941685287969437,
      "grad_norm": 0.19091794041878155,
      "learning_rate": 0.00016558867243375314,
      "loss": 0.9847,
      "step": 4081
    },
    {
      "epoch": 0.29424061125928064,
      "grad_norm": 1.1327167171369512,
      "learning_rate": 0.0001655710468315332,
      "loss": 1.1519,
      "step": 4082
    },
    {
      "epoch": 0.29431269372161756,
      "grad_norm": 0.11325200780101352,
      "learning_rate": 0.00016555341765510463,
      "loss": 1.0912,
      "step": 4083
    },
    {
      "epoch": 0.2943847761839544,
      "grad_norm": 0.16842404561763025,
      "learning_rate": 0.00016553578490542832,
      "loss": 1.0623,
      "step": 4084
    },
    {
      "epoch": 0.29445685864629134,
      "grad_norm": 0.12311904652261194,
      "learning_rate": 0.00016551814858346543,
      "loss": 1.0507,
      "step": 4085
    },
    {
      "epoch": 0.29452894110862826,
      "grad_norm": 0.1311318033791874,
      "learning_rate": 0.0001655005086901773,
      "loss": 1.1369,
      "step": 4086
    },
    {
      "epoch": 0.2946010235709652,
      "grad_norm": 0.20554550776210845,
      "learning_rate": 0.00016548286522652546,
      "loss": 1.0958,
      "step": 4087
    },
    {
      "epoch": 0.2946731060333021,
      "grad_norm": 0.12256300289031061,
      "learning_rate": 0.00016546521819347167,
      "loss": 1.1151,
      "step": 4088
    },
    {
      "epoch": 0.294745188495639,
      "grad_norm": 0.18040219002392152,
      "learning_rate": 0.00016544756759197778,
      "loss": 1.1117,
      "step": 4089
    },
    {
      "epoch": 0.29481727095797594,
      "grad_norm": 0.20724653263754805,
      "learning_rate": 0.00016542991342300597,
      "loss": 1.0308,
      "step": 4090
    },
    {
      "epoch": 0.29488935342031286,
      "grad_norm": 0.11269222225751079,
      "learning_rate": 0.00016541225568751853,
      "loss": 1.0166,
      "step": 4091
    },
    {
      "epoch": 0.2949614358826497,
      "grad_norm": 0.6158069092811399,
      "learning_rate": 0.00016539459438647794,
      "loss": 1.0858,
      "step": 4092
    },
    {
      "epoch": 0.29503351834498664,
      "grad_norm": 0.19500990376466665,
      "learning_rate": 0.0001653769295208469,
      "loss": 1.0762,
      "step": 4093
    },
    {
      "epoch": 0.29510560080732356,
      "grad_norm": 0.19479660782548833,
      "learning_rate": 0.00016535926109158836,
      "loss": 1.1249,
      "step": 4094
    },
    {
      "epoch": 0.2951776832696605,
      "grad_norm": 0.1132722707001632,
      "learning_rate": 0.00016534158909966537,
      "loss": 1.0544,
      "step": 4095
    },
    {
      "epoch": 0.2952497657319974,
      "grad_norm": 0.5788432065647783,
      "learning_rate": 0.00016532391354604119,
      "loss": 1.0408,
      "step": 4096
    },
    {
      "epoch": 0.2953218481943343,
      "grad_norm": 0.24495466347353564,
      "learning_rate": 0.00016530623443167933,
      "loss": 1.0976,
      "step": 4097
    },
    {
      "epoch": 0.29539393065667124,
      "grad_norm": 0.12606232005335208,
      "learning_rate": 0.00016528855175754344,
      "loss": 1.0647,
      "step": 4098
    },
    {
      "epoch": 0.29546601311900816,
      "grad_norm": 0.6510270849369092,
      "learning_rate": 0.0001652708655245974,
      "loss": 1.0411,
      "step": 4099
    },
    {
      "epoch": 0.2955380955813451,
      "grad_norm": 0.12694719172188879,
      "learning_rate": 0.00016525317573380525,
      "loss": 1.0759,
      "step": 4100
    },
    {
      "epoch": 0.29561017804368195,
      "grad_norm": 0.20288584660064635,
      "learning_rate": 0.00016523548238613127,
      "loss": 1.0418,
      "step": 4101
    },
    {
      "epoch": 0.29568226050601887,
      "grad_norm": 0.20849862717208675,
      "learning_rate": 0.00016521778548253985,
      "loss": 1.0052,
      "step": 4102
    },
    {
      "epoch": 0.2957543429683558,
      "grad_norm": 0.1599074377590435,
      "learning_rate": 0.00016520008502399571,
      "loss": 1.1231,
      "step": 4103
    },
    {
      "epoch": 0.2958264254306927,
      "grad_norm": 0.141568871858861,
      "learning_rate": 0.0001651823810114636,
      "loss": 1.1212,
      "step": 4104
    },
    {
      "epoch": 0.2958985078930296,
      "grad_norm": 0.16789237289035516,
      "learning_rate": 0.0001651646734459086,
      "loss": 1.0914,
      "step": 4105
    },
    {
      "epoch": 0.29597059035536655,
      "grad_norm": 0.5670608007666929,
      "learning_rate": 0.00016514696232829593,
      "loss": 1.1272,
      "step": 4106
    },
    {
      "epoch": 0.29604267281770347,
      "grad_norm": 0.5247443930584098,
      "learning_rate": 0.000165129247659591,
      "loss": 1.1089,
      "step": 4107
    },
    {
      "epoch": 0.2961147552800404,
      "grad_norm": 0.150898053084417,
      "learning_rate": 0.0001651115294407594,
      "loss": 1.0519,
      "step": 4108
    },
    {
      "epoch": 0.2961868377423773,
      "grad_norm": 0.14006300174454817,
      "learning_rate": 0.0001650938076727669,
      "loss": 1.0865,
      "step": 4109
    },
    {
      "epoch": 0.29625892020471417,
      "grad_norm": 0.1735956930666292,
      "learning_rate": 0.00016507608235657962,
      "loss": 1.0854,
      "step": 4110
    },
    {
      "epoch": 0.2963310026670511,
      "grad_norm": 0.2107858080986491,
      "learning_rate": 0.00016505835349316358,
      "loss": 1.0341,
      "step": 4111
    },
    {
      "epoch": 0.296403085129388,
      "grad_norm": 0.22915803230055287,
      "learning_rate": 0.0001650406210834853,
      "loss": 1.0475,
      "step": 4112
    },
    {
      "epoch": 0.29647516759172493,
      "grad_norm": 0.183582033454671,
      "learning_rate": 0.00016502288512851125,
      "loss": 1.0742,
      "step": 4113
    },
    {
      "epoch": 0.29654725005406185,
      "grad_norm": 0.12769143544610576,
      "learning_rate": 0.00016500514562920831,
      "loss": 1.0514,
      "step": 4114
    },
    {
      "epoch": 0.29661933251639877,
      "grad_norm": 0.13829468145247195,
      "learning_rate": 0.00016498740258654335,
      "loss": 1.0778,
      "step": 4115
    },
    {
      "epoch": 0.2966914149787357,
      "grad_norm": 0.13066960657501858,
      "learning_rate": 0.00016496965600148355,
      "loss": 1.0734,
      "step": 4116
    },
    {
      "epoch": 0.2967634974410726,
      "grad_norm": 0.19966489708537952,
      "learning_rate": 0.00016495190587499626,
      "loss": 1.0826,
      "step": 4117
    },
    {
      "epoch": 0.2968355799034095,
      "grad_norm": 0.5701623658313606,
      "learning_rate": 0.00016493415220804905,
      "loss": 0.9997,
      "step": 4118
    },
    {
      "epoch": 0.2969076623657464,
      "grad_norm": 0.12932992220566544,
      "learning_rate": 0.00016491639500160958,
      "loss": 1.0345,
      "step": 4119
    },
    {
      "epoch": 0.2969797448280833,
      "grad_norm": 0.7373183541742352,
      "learning_rate": 0.00016489863425664587,
      "loss": 1.0471,
      "step": 4120
    },
    {
      "epoch": 0.29705182729042023,
      "grad_norm": 0.15436830595328782,
      "learning_rate": 0.00016488086997412598,
      "loss": 1.1038,
      "step": 4121
    },
    {
      "epoch": 0.29712390975275715,
      "grad_norm": 0.18535947050180152,
      "learning_rate": 0.00016486310215501823,
      "loss": 1.0651,
      "step": 4122
    },
    {
      "epoch": 0.2971959922150941,
      "grad_norm": 0.13272869107457871,
      "learning_rate": 0.00016484533080029113,
      "loss": 1.0566,
      "step": 4123
    },
    {
      "epoch": 0.297268074677431,
      "grad_norm": 0.14446234296727398,
      "learning_rate": 0.00016482755591091337,
      "loss": 1.0265,
      "step": 4124
    },
    {
      "epoch": 0.2973401571397679,
      "grad_norm": 0.1389191111266714,
      "learning_rate": 0.00016480977748785386,
      "loss": 1.0573,
      "step": 4125
    },
    {
      "epoch": 0.29741223960210483,
      "grad_norm": 0.2001228029108409,
      "learning_rate": 0.0001647919955320817,
      "loss": 1.0488,
      "step": 4126
    },
    {
      "epoch": 0.2974843220644417,
      "grad_norm": 0.33999608985195917,
      "learning_rate": 0.00016477421004456607,
      "loss": 1.027,
      "step": 4127
    },
    {
      "epoch": 0.2975564045267786,
      "grad_norm": 0.2847811378682866,
      "learning_rate": 0.00016475642102627657,
      "loss": 1.0739,
      "step": 4128
    },
    {
      "epoch": 0.29762848698911554,
      "grad_norm": 1.2647752482660806,
      "learning_rate": 0.00016473862847818277,
      "loss": 1.1092,
      "step": 4129
    },
    {
      "epoch": 0.29770056945145246,
      "grad_norm": 0.12756021465905637,
      "learning_rate": 0.00016472083240125454,
      "loss": 1.1028,
      "step": 4130
    },
    {
      "epoch": 0.2977726519137894,
      "grad_norm": 0.1352362641523953,
      "learning_rate": 0.00016470303279646196,
      "loss": 1.0386,
      "step": 4131
    },
    {
      "epoch": 0.2978447343761263,
      "grad_norm": 0.15101548954767163,
      "learning_rate": 0.00016468522966477523,
      "loss": 1.1496,
      "step": 4132
    },
    {
      "epoch": 0.2979168168384632,
      "grad_norm": 0.12014942591723116,
      "learning_rate": 0.0001646674230071648,
      "loss": 1.0233,
      "step": 4133
    },
    {
      "epoch": 0.29798889930080014,
      "grad_norm": 0.131769370496643,
      "learning_rate": 0.00016464961282460126,
      "loss": 1.0858,
      "step": 4134
    },
    {
      "epoch": 0.29806098176313706,
      "grad_norm": 0.1691689219581204,
      "learning_rate": 0.0001646317991180555,
      "loss": 1.1017,
      "step": 4135
    },
    {
      "epoch": 0.2981330642254739,
      "grad_norm": 1.4742865550277,
      "learning_rate": 0.0001646139818884984,
      "loss": 0.9678,
      "step": 4136
    },
    {
      "epoch": 0.29820514668781084,
      "grad_norm": 0.11877504780023176,
      "learning_rate": 0.0001645961611369013,
      "loss": 1.0485,
      "step": 4137
    },
    {
      "epoch": 0.29827722915014776,
      "grad_norm": 0.1284578134494099,
      "learning_rate": 0.0001645783368642355,
      "loss": 1.0026,
      "step": 4138
    },
    {
      "epoch": 0.2983493116124847,
      "grad_norm": 0.11946539267056748,
      "learning_rate": 0.0001645605090714726,
      "loss": 1.0411,
      "step": 4139
    },
    {
      "epoch": 0.2984213940748216,
      "grad_norm": 0.1824587124465221,
      "learning_rate": 0.00016454267775958437,
      "loss": 1.1503,
      "step": 4140
    },
    {
      "epoch": 0.2984934765371585,
      "grad_norm": 0.11531951000239238,
      "learning_rate": 0.00016452484292954283,
      "loss": 1.1596,
      "step": 4141
    },
    {
      "epoch": 0.29856555899949544,
      "grad_norm": 0.1315215728151176,
      "learning_rate": 0.00016450700458232008,
      "loss": 1.0934,
      "step": 4142
    },
    {
      "epoch": 0.29863764146183236,
      "grad_norm": 0.3236935764809063,
      "learning_rate": 0.00016448916271888847,
      "loss": 1.1119,
      "step": 4143
    },
    {
      "epoch": 0.2987097239241692,
      "grad_norm": 0.12967813074083484,
      "learning_rate": 0.00016447131734022055,
      "loss": 1.0433,
      "step": 4144
    },
    {
      "epoch": 0.29878180638650614,
      "grad_norm": 0.14719311914307814,
      "learning_rate": 0.00016445346844728905,
      "loss": 1.1026,
      "step": 4145
    },
    {
      "epoch": 0.29885388884884306,
      "grad_norm": 0.1815357118696479,
      "learning_rate": 0.00016443561604106695,
      "loss": 1.0259,
      "step": 4146
    },
    {
      "epoch": 0.29892597131118,
      "grad_norm": 0.19170140080970752,
      "learning_rate": 0.00016441776012252727,
      "loss": 1.0989,
      "step": 4147
    },
    {
      "epoch": 0.2989980537735169,
      "grad_norm": 0.13977889822774245,
      "learning_rate": 0.00016439990069264338,
      "loss": 1.0533,
      "step": 4148
    },
    {
      "epoch": 0.2990701362358538,
      "grad_norm": 0.12783333561044086,
      "learning_rate": 0.00016438203775238875,
      "loss": 1.0036,
      "step": 4149
    },
    {
      "epoch": 0.29914221869819074,
      "grad_norm": 1.0025645231734188,
      "learning_rate": 0.00016436417130273712,
      "loss": 1.0691,
      "step": 4150
    },
    {
      "epoch": 0.29921430116052766,
      "grad_norm": 0.2474030899060875,
      "learning_rate": 0.0001643463013446623,
      "loss": 1.1238,
      "step": 4151
    },
    {
      "epoch": 0.2992863836228646,
      "grad_norm": 0.2153195892426891,
      "learning_rate": 0.00016432842787913841,
      "loss": 1.048,
      "step": 4152
    },
    {
      "epoch": 0.29935846608520145,
      "grad_norm": 0.14855103431179728,
      "learning_rate": 0.00016431055090713974,
      "loss": 1.0873,
      "step": 4153
    },
    {
      "epoch": 0.29943054854753837,
      "grad_norm": 0.13950690661847867,
      "learning_rate": 0.00016429267042964067,
      "loss": 1.0678,
      "step": 4154
    },
    {
      "epoch": 0.2995026310098753,
      "grad_norm": 0.15096191224848007,
      "learning_rate": 0.00016427478644761586,
      "loss": 1.0468,
      "step": 4155
    },
    {
      "epoch": 0.2995747134722122,
      "grad_norm": 0.1651578478399883,
      "learning_rate": 0.00016425689896204025,
      "loss": 1.061,
      "step": 4156
    },
    {
      "epoch": 0.2996467959345491,
      "grad_norm": 0.16724546399222381,
      "learning_rate": 0.0001642390079738887,
      "loss": 1.0432,
      "step": 4157
    },
    {
      "epoch": 0.29971887839688605,
      "grad_norm": 0.13017597400562952,
      "learning_rate": 0.00016422111348413657,
      "loss": 1.1041,
      "step": 4158
    },
    {
      "epoch": 0.29979096085922297,
      "grad_norm": 0.24719766698530776,
      "learning_rate": 0.00016420321549375918,
      "loss": 1.065,
      "step": 4159
    },
    {
      "epoch": 0.2998630433215599,
      "grad_norm": 0.16523383204519113,
      "learning_rate": 0.0001641853140037322,
      "loss": 1.0819,
      "step": 4160
    },
    {
      "epoch": 0.2999351257838968,
      "grad_norm": 0.12851630242471457,
      "learning_rate": 0.00016416740901503135,
      "loss": 1.0016,
      "step": 4161
    },
    {
      "epoch": 0.30000720824623367,
      "grad_norm": 0.19299718500478327,
      "learning_rate": 0.0001641495005286327,
      "loss": 1.1298,
      "step": 4162
    },
    {
      "epoch": 0.3000792907085706,
      "grad_norm": 0.33358080785175115,
      "learning_rate": 0.00016413158854551232,
      "loss": 1.0724,
      "step": 4163
    },
    {
      "epoch": 0.3001513731709075,
      "grad_norm": 0.3710752519185079,
      "learning_rate": 0.00016411367306664668,
      "loss": 1.0795,
      "step": 4164
    },
    {
      "epoch": 0.30022345563324443,
      "grad_norm": 0.17487354542680175,
      "learning_rate": 0.00016409575409301227,
      "loss": 0.9867,
      "step": 4165
    },
    {
      "epoch": 0.30029553809558135,
      "grad_norm": 1.1137387586963752,
      "learning_rate": 0.0001640778316255858,
      "loss": 1.0221,
      "step": 4166
    },
    {
      "epoch": 0.30036762055791827,
      "grad_norm": 0.16015224918900162,
      "learning_rate": 0.0001640599056653443,
      "loss": 1.0979,
      "step": 4167
    },
    {
      "epoch": 0.3004397030202552,
      "grad_norm": 0.15486528480850714,
      "learning_rate": 0.00016404197621326482,
      "loss": 1.0799,
      "step": 4168
    },
    {
      "epoch": 0.3005117854825921,
      "grad_norm": 0.13760998086863263,
      "learning_rate": 0.0001640240432703247,
      "loss": 1.1593,
      "step": 4169
    },
    {
      "epoch": 0.300583867944929,
      "grad_norm": 0.22553095179111923,
      "learning_rate": 0.00016400610683750143,
      "loss": 1.0322,
      "step": 4170
    },
    {
      "epoch": 0.3006559504072659,
      "grad_norm": 0.1130088032206278,
      "learning_rate": 0.0001639881669157728,
      "loss": 1.1142,
      "step": 4171
    },
    {
      "epoch": 0.3007280328696028,
      "grad_norm": 0.1212031293169529,
      "learning_rate": 0.00016397022350611652,
      "loss": 1.0636,
      "step": 4172
    },
    {
      "epoch": 0.30080011533193973,
      "grad_norm": 0.13021878467956177,
      "learning_rate": 0.00016395227660951083,
      "loss": 1.0752,
      "step": 4173
    },
    {
      "epoch": 0.30087219779427665,
      "grad_norm": 0.24228743873204425,
      "learning_rate": 0.0001639343262269339,
      "loss": 1.0225,
      "step": 4174
    },
    {
      "epoch": 0.30094428025661357,
      "grad_norm": 0.13604979759170077,
      "learning_rate": 0.00016391637235936424,
      "loss": 1.1457,
      "step": 4175
    },
    {
      "epoch": 0.3010163627189505,
      "grad_norm": 0.11986802916918703,
      "learning_rate": 0.00016389841500778047,
      "loss": 0.9573,
      "step": 4176
    },
    {
      "epoch": 0.3010884451812874,
      "grad_norm": 0.18639792014441972,
      "learning_rate": 0.0001638804541731614,
      "loss": 1.078,
      "step": 4177
    },
    {
      "epoch": 0.30116052764362433,
      "grad_norm": 0.16097010405170228,
      "learning_rate": 0.00016386248985648613,
      "loss": 1.0789,
      "step": 4178
    },
    {
      "epoch": 0.3012326101059612,
      "grad_norm": 0.11493054572445788,
      "learning_rate": 0.00016384452205873377,
      "loss": 1.0875,
      "step": 4179
    },
    {
      "epoch": 0.3013046925682981,
      "grad_norm": 0.1400221550007291,
      "learning_rate": 0.00016382655078088387,
      "loss": 1.0798,
      "step": 4180
    },
    {
      "epoch": 0.30137677503063504,
      "grad_norm": 0.13038506504068545,
      "learning_rate": 0.0001638085760239159,
      "loss": 1.058,
      "step": 4181
    },
    {
      "epoch": 0.30144885749297196,
      "grad_norm": 0.14248690545336287,
      "learning_rate": 0.00016379059778880967,
      "loss": 1.0817,
      "step": 4182
    },
    {
      "epoch": 0.3015209399553089,
      "grad_norm": 0.1683581688020347,
      "learning_rate": 0.00016377261607654518,
      "loss": 1.0953,
      "step": 4183
    },
    {
      "epoch": 0.3015930224176458,
      "grad_norm": 0.1209667523531658,
      "learning_rate": 0.00016375463088810264,
      "loss": 1.008,
      "step": 4184
    },
    {
      "epoch": 0.3016651048799827,
      "grad_norm": 0.15961524187876205,
      "learning_rate": 0.0001637366422244623,
      "loss": 1.1163,
      "step": 4185
    },
    {
      "epoch": 0.30173718734231963,
      "grad_norm": 0.1226341184929483,
      "learning_rate": 0.00016371865008660475,
      "loss": 1.1368,
      "step": 4186
    },
    {
      "epoch": 0.30180926980465655,
      "grad_norm": 0.12724058948633005,
      "learning_rate": 0.00016370065447551078,
      "loss": 1.0715,
      "step": 4187
    },
    {
      "epoch": 0.3018813522669934,
      "grad_norm": 0.11910887517126223,
      "learning_rate": 0.0001636826553921612,
      "loss": 1.0619,
      "step": 4188
    },
    {
      "epoch": 0.30195343472933034,
      "grad_norm": 0.13984431148611645,
      "learning_rate": 0.00016366465283753723,
      "loss": 0.9976,
      "step": 4189
    },
    {
      "epoch": 0.30202551719166726,
      "grad_norm": 0.12973347845443173,
      "learning_rate": 0.0001636466468126201,
      "loss": 0.991,
      "step": 4190
    },
    {
      "epoch": 0.3020975996540042,
      "grad_norm": 0.1200337142561169,
      "learning_rate": 0.00016362863731839128,
      "loss": 1.1433,
      "step": 4191
    },
    {
      "epoch": 0.3021696821163411,
      "grad_norm": 0.13457217927607096,
      "learning_rate": 0.00016361062435583254,
      "loss": 1.0509,
      "step": 4192
    },
    {
      "epoch": 0.302241764578678,
      "grad_norm": 0.22397368458951583,
      "learning_rate": 0.00016359260792592567,
      "loss": 1.0719,
      "step": 4193
    },
    {
      "epoch": 0.30231384704101494,
      "grad_norm": 0.21286273332471217,
      "learning_rate": 0.00016357458802965276,
      "loss": 1.0995,
      "step": 4194
    },
    {
      "epoch": 0.30238592950335186,
      "grad_norm": 0.14265068258688246,
      "learning_rate": 0.00016355656466799604,
      "loss": 1.0603,
      "step": 4195
    },
    {
      "epoch": 0.3024580119656887,
      "grad_norm": 0.14588789067301108,
      "learning_rate": 0.00016353853784193797,
      "loss": 1.0255,
      "step": 4196
    },
    {
      "epoch": 0.30253009442802564,
      "grad_norm": 0.21577904559240796,
      "learning_rate": 0.00016352050755246117,
      "loss": 1.0263,
      "step": 4197
    },
    {
      "epoch": 0.30260217689036256,
      "grad_norm": 0.12015467534163084,
      "learning_rate": 0.0001635024738005484,
      "loss": 0.9589,
      "step": 4198
    },
    {
      "epoch": 0.3026742593526995,
      "grad_norm": 0.15792782233417427,
      "learning_rate": 0.0001634844365871827,
      "loss": 1.0369,
      "step": 4199
    },
    {
      "epoch": 0.3027463418150364,
      "grad_norm": 0.14968913518940596,
      "learning_rate": 0.0001634663959133473,
      "loss": 1.0547,
      "step": 4200
    },
    {
      "epoch": 0.3028184242773733,
      "grad_norm": 0.13080430292681328,
      "learning_rate": 0.00016344835178002548,
      "loss": 1.0787,
      "step": 4201
    },
    {
      "epoch": 0.30289050673971024,
      "grad_norm": 0.11359249436912129,
      "learning_rate": 0.0001634303041882009,
      "loss": 1.0702,
      "step": 4202
    },
    {
      "epoch": 0.30296258920204716,
      "grad_norm": 0.1340785612229693,
      "learning_rate": 0.00016341225313885727,
      "loss": 1.071,
      "step": 4203
    },
    {
      "epoch": 0.3030346716643841,
      "grad_norm": 0.13526731311776038,
      "learning_rate": 0.00016339419863297853,
      "loss": 1.1376,
      "step": 4204
    },
    {
      "epoch": 0.30310675412672095,
      "grad_norm": 0.25321576401474827,
      "learning_rate": 0.00016337614067154883,
      "loss": 1.0503,
      "step": 4205
    },
    {
      "epoch": 0.30317883658905787,
      "grad_norm": 0.21997056497129083,
      "learning_rate": 0.00016335807925555248,
      "loss": 1.0859,
      "step": 4206
    },
    {
      "epoch": 0.3032509190513948,
      "grad_norm": 0.18545664260915773,
      "learning_rate": 0.000163340014385974,
      "loss": 1.1835,
      "step": 4207
    },
    {
      "epoch": 0.3033230015137317,
      "grad_norm": 0.12613167381017526,
      "learning_rate": 0.0001633219460637981,
      "loss": 1.0814,
      "step": 4208
    },
    {
      "epoch": 0.3033950839760686,
      "grad_norm": 0.12401979754680267,
      "learning_rate": 0.00016330387429000962,
      "loss": 0.9843,
      "step": 4209
    },
    {
      "epoch": 0.30346716643840554,
      "grad_norm": 0.13596478748215626,
      "learning_rate": 0.00016328579906559366,
      "loss": 1.0533,
      "step": 4210
    },
    {
      "epoch": 0.30353924890074246,
      "grad_norm": 0.1493317328425291,
      "learning_rate": 0.0001632677203915355,
      "loss": 1.0091,
      "step": 4211
    },
    {
      "epoch": 0.3036113313630794,
      "grad_norm": 0.17149433955319962,
      "learning_rate": 0.00016324963826882055,
      "loss": 1.0445,
      "step": 4212
    },
    {
      "epoch": 0.30368341382541625,
      "grad_norm": 0.18995178836012563,
      "learning_rate": 0.00016323155269843447,
      "loss": 1.0041,
      "step": 4213
    },
    {
      "epoch": 0.30375549628775317,
      "grad_norm": 0.1475026787340501,
      "learning_rate": 0.00016321346368136305,
      "loss": 1.0559,
      "step": 4214
    },
    {
      "epoch": 0.3038275787500901,
      "grad_norm": 0.13988115977966245,
      "learning_rate": 0.0001631953712185924,
      "loss": 1.065,
      "step": 4215
    },
    {
      "epoch": 0.303899661212427,
      "grad_norm": 0.13804589258971048,
      "learning_rate": 0.0001631772753111086,
      "loss": 0.9458,
      "step": 4216
    },
    {
      "epoch": 0.30397174367476393,
      "grad_norm": 0.14106808856056197,
      "learning_rate": 0.00016315917595989814,
      "loss": 1.0414,
      "step": 4217
    },
    {
      "epoch": 0.30404382613710085,
      "grad_norm": 0.17511064667171328,
      "learning_rate": 0.00016314107316594753,
      "loss": 1.111,
      "step": 4218
    },
    {
      "epoch": 0.30411590859943777,
      "grad_norm": 0.14376291508054526,
      "learning_rate": 0.00016312296693024356,
      "loss": 1.03,
      "step": 4219
    },
    {
      "epoch": 0.3041879910617747,
      "grad_norm": 0.14779088259766404,
      "learning_rate": 0.0001631048572537732,
      "loss": 1.0507,
      "step": 4220
    },
    {
      "epoch": 0.3042600735241116,
      "grad_norm": 0.1575087822706677,
      "learning_rate": 0.00016308674413752352,
      "loss": 1.0401,
      "step": 4221
    },
    {
      "epoch": 0.30433215598644847,
      "grad_norm": 0.12325585685355112,
      "learning_rate": 0.00016306862758248193,
      "loss": 1.0948,
      "step": 4222
    },
    {
      "epoch": 0.3044042384487854,
      "grad_norm": 0.1440252999204568,
      "learning_rate": 0.00016305050758963588,
      "loss": 1.0782,
      "step": 4223
    },
    {
      "epoch": 0.3044763209111223,
      "grad_norm": 0.3803221685311305,
      "learning_rate": 0.00016303238415997314,
      "loss": 1.122,
      "step": 4224
    },
    {
      "epoch": 0.30454840337345923,
      "grad_norm": 0.12858630465892834,
      "learning_rate": 0.0001630142572944815,
      "loss": 1.0547,
      "step": 4225
    },
    {
      "epoch": 0.30462048583579615,
      "grad_norm": 0.11666153822229657,
      "learning_rate": 0.00016299612699414916,
      "loss": 1.0544,
      "step": 4226
    },
    {
      "epoch": 0.30469256829813307,
      "grad_norm": 0.1289581082543608,
      "learning_rate": 0.0001629779932599643,
      "loss": 1.0667,
      "step": 4227
    },
    {
      "epoch": 0.30476465076047,
      "grad_norm": 0.14705105804381605,
      "learning_rate": 0.00016295985609291534,
      "loss": 1.0015,
      "step": 4228
    },
    {
      "epoch": 0.3048367332228069,
      "grad_norm": 0.1339211189203788,
      "learning_rate": 0.00016294171549399104,
      "loss": 1.0325,
      "step": 4229
    },
    {
      "epoch": 0.30490881568514383,
      "grad_norm": 0.15154988996976426,
      "learning_rate": 0.0001629235714641801,
      "loss": 1.1348,
      "step": 4230
    },
    {
      "epoch": 0.3049808981474807,
      "grad_norm": 0.13610167255919348,
      "learning_rate": 0.00016290542400447164,
      "loss": 1.0862,
      "step": 4231
    },
    {
      "epoch": 0.3050529806098176,
      "grad_norm": 0.24292918545452508,
      "learning_rate": 0.00016288727311585473,
      "loss": 1.1099,
      "step": 4232
    },
    {
      "epoch": 0.30512506307215453,
      "grad_norm": 0.14047169596654513,
      "learning_rate": 0.0001628691187993189,
      "loss": 1.0121,
      "step": 4233
    },
    {
      "epoch": 0.30519714553449145,
      "grad_norm": 0.16215584517183293,
      "learning_rate": 0.00016285096105585363,
      "loss": 1.0777,
      "step": 4234
    },
    {
      "epoch": 0.3052692279968284,
      "grad_norm": 0.11436317756722135,
      "learning_rate": 0.00016283279988644868,
      "loss": 1.0414,
      "step": 4235
    },
    {
      "epoch": 0.3053413104591653,
      "grad_norm": 0.11923784907673743,
      "learning_rate": 0.00016281463529209405,
      "loss": 1.129,
      "step": 4236
    },
    {
      "epoch": 0.3054133929215022,
      "grad_norm": 0.11800587640805126,
      "learning_rate": 0.00016279646727377983,
      "loss": 1.0496,
      "step": 4237
    },
    {
      "epoch": 0.30548547538383913,
      "grad_norm": 0.544216531025566,
      "learning_rate": 0.00016277829583249634,
      "loss": 1.0638,
      "step": 4238
    },
    {
      "epoch": 0.305557557846176,
      "grad_norm": 0.11504778538860515,
      "learning_rate": 0.0001627601209692341,
      "loss": 1.014,
      "step": 4239
    },
    {
      "epoch": 0.3056296403085129,
      "grad_norm": 0.1610100512509126,
      "learning_rate": 0.0001627419426849838,
      "loss": 1.0928,
      "step": 4240
    },
    {
      "epoch": 0.30570172277084984,
      "grad_norm": 0.10998377892942467,
      "learning_rate": 0.0001627237609807363,
      "loss": 0.987,
      "step": 4241
    },
    {
      "epoch": 0.30577380523318676,
      "grad_norm": 0.5433083893556441,
      "learning_rate": 0.0001627055758574827,
      "loss": 1.0885,
      "step": 4242
    },
    {
      "epoch": 0.3058458876955237,
      "grad_norm": 0.15156276416779235,
      "learning_rate": 0.00016268738731621424,
      "loss": 1.0496,
      "step": 4243
    },
    {
      "epoch": 0.3059179701578606,
      "grad_norm": 0.13027923892856436,
      "learning_rate": 0.00016266919535792238,
      "loss": 1.0376,
      "step": 4244
    },
    {
      "epoch": 0.3059900526201975,
      "grad_norm": 0.14450412424094974,
      "learning_rate": 0.00016265099998359866,
      "loss": 1.1455,
      "step": 4245
    },
    {
      "epoch": 0.30606213508253444,
      "grad_norm": 0.20562402581188563,
      "learning_rate": 0.000162632801194235,
      "loss": 1.0748,
      "step": 4246
    },
    {
      "epoch": 0.30613421754487136,
      "grad_norm": 0.19060606042358416,
      "learning_rate": 0.0001626145989908233,
      "loss": 1.0336,
      "step": 4247
    },
    {
      "epoch": 0.3062063000072082,
      "grad_norm": 0.14556281072236976,
      "learning_rate": 0.00016259639337435583,
      "loss": 1.0951,
      "step": 4248
    },
    {
      "epoch": 0.30627838246954514,
      "grad_norm": 0.11207329784019127,
      "learning_rate": 0.0001625781843458249,
      "loss": 1.1458,
      "step": 4249
    },
    {
      "epoch": 0.30635046493188206,
      "grad_norm": 0.1872818387675678,
      "learning_rate": 0.0001625599719062231,
      "loss": 1.089,
      "step": 4250
    },
    {
      "epoch": 0.306422547394219,
      "grad_norm": 0.11260759451394572,
      "learning_rate": 0.00016254175605654312,
      "loss": 1.052,
      "step": 4251
    },
    {
      "epoch": 0.3064946298565559,
      "grad_norm": 0.15798777191137,
      "learning_rate": 0.00016252353679777794,
      "loss": 1.0746,
      "step": 4252
    },
    {
      "epoch": 0.3065667123188928,
      "grad_norm": 0.2831495441729306,
      "learning_rate": 0.00016250531413092065,
      "loss": 0.9991,
      "step": 4253
    },
    {
      "epoch": 0.30663879478122974,
      "grad_norm": 0.2659553019905205,
      "learning_rate": 0.00016248708805696456,
      "loss": 1.1309,
      "step": 4254
    },
    {
      "epoch": 0.30671087724356666,
      "grad_norm": 0.32646018469662685,
      "learning_rate": 0.00016246885857690315,
      "loss": 1.0843,
      "step": 4255
    },
    {
      "epoch": 0.3067829597059036,
      "grad_norm": 0.13696581911858555,
      "learning_rate": 0.00016245062569173005,
      "loss": 1.0755,
      "step": 4256
    },
    {
      "epoch": 0.30685504216824044,
      "grad_norm": 0.12882598639940268,
      "learning_rate": 0.0001624323894024392,
      "loss": 1.0411,
      "step": 4257
    },
    {
      "epoch": 0.30692712463057736,
      "grad_norm": 0.5966848708635477,
      "learning_rate": 0.00016241414971002454,
      "loss": 1.0493,
      "step": 4258
    },
    {
      "epoch": 0.3069992070929143,
      "grad_norm": 0.3658645159755085,
      "learning_rate": 0.00016239590661548035,
      "loss": 1.0853,
      "step": 4259
    },
    {
      "epoch": 0.3070712895552512,
      "grad_norm": 0.149031746944972,
      "learning_rate": 0.0001623776601198011,
      "loss": 1.0409,
      "step": 4260
    },
    {
      "epoch": 0.3071433720175881,
      "grad_norm": 0.14504440904114957,
      "learning_rate": 0.0001623594102239813,
      "loss": 1.0832,
      "step": 4261
    },
    {
      "epoch": 0.30721545447992504,
      "grad_norm": 0.13093470886625888,
      "learning_rate": 0.00016234115692901572,
      "loss": 1.0427,
      "step": 4262
    },
    {
      "epoch": 0.30728753694226196,
      "grad_norm": 0.12605200955168375,
      "learning_rate": 0.0001623229002358994,
      "loss": 1.0878,
      "step": 4263
    },
    {
      "epoch": 0.3073596194045989,
      "grad_norm": 0.1786989814489302,
      "learning_rate": 0.00016230464014562747,
      "loss": 1.0588,
      "step": 4264
    },
    {
      "epoch": 0.30743170186693575,
      "grad_norm": 0.1140011371226909,
      "learning_rate": 0.00016228637665919525,
      "loss": 1.0819,
      "step": 4265
    },
    {
      "epoch": 0.30750378432927267,
      "grad_norm": 0.11903447687955986,
      "learning_rate": 0.00016226810977759825,
      "loss": 1.0519,
      "step": 4266
    },
    {
      "epoch": 0.3075758667916096,
      "grad_norm": 0.16007329298671116,
      "learning_rate": 0.00016224983950183222,
      "loss": 1.0654,
      "step": 4267
    },
    {
      "epoch": 0.3076479492539465,
      "grad_norm": 0.15656495624763492,
      "learning_rate": 0.00016223156583289306,
      "loss": 1.072,
      "step": 4268
    },
    {
      "epoch": 0.3077200317162834,
      "grad_norm": 0.16048742389952947,
      "learning_rate": 0.00016221328877177678,
      "loss": 1.0923,
      "step": 4269
    },
    {
      "epoch": 0.30779211417862035,
      "grad_norm": 0.11022755102943915,
      "learning_rate": 0.0001621950083194797,
      "loss": 1.0992,
      "step": 4270
    },
    {
      "epoch": 0.30786419664095727,
      "grad_norm": 0.3105403663777658,
      "learning_rate": 0.00016217672447699824,
      "loss": 1.0837,
      "step": 4271
    },
    {
      "epoch": 0.3079362791032942,
      "grad_norm": 0.38039693412272924,
      "learning_rate": 0.00016215843724532908,
      "loss": 1.0074,
      "step": 4272
    },
    {
      "epoch": 0.3080083615656311,
      "grad_norm": 0.127004027789808,
      "learning_rate": 0.00016214014662546897,
      "loss": 1.035,
      "step": 4273
    },
    {
      "epoch": 0.30808044402796797,
      "grad_norm": 0.1660148369965778,
      "learning_rate": 0.00016212185261841499,
      "loss": 1.058,
      "step": 4274
    },
    {
      "epoch": 0.3081525264903049,
      "grad_norm": 0.14867696033629718,
      "learning_rate": 0.00016210355522516424,
      "loss": 1.0459,
      "step": 4275
    },
    {
      "epoch": 0.3082246089526418,
      "grad_norm": 0.11617218965507899,
      "learning_rate": 0.00016208525444671415,
      "loss": 1.0794,
      "step": 4276
    },
    {
      "epoch": 0.30829669141497873,
      "grad_norm": 0.1742415989042181,
      "learning_rate": 0.00016206695028406229,
      "loss": 1.0155,
      "step": 4277
    },
    {
      "epoch": 0.30836877387731565,
      "grad_norm": 0.15703900269384496,
      "learning_rate": 0.00016204864273820633,
      "loss": 1.0924,
      "step": 4278
    },
    {
      "epoch": 0.30844085633965257,
      "grad_norm": 0.24373033548879644,
      "learning_rate": 0.00016203033181014427,
      "loss": 1.1054,
      "step": 4279
    },
    {
      "epoch": 0.3085129388019895,
      "grad_norm": 0.13076963875625133,
      "learning_rate": 0.00016201201750087418,
      "loss": 1.0474,
      "step": 4280
    },
    {
      "epoch": 0.3085850212643264,
      "grad_norm": 0.32612954461083954,
      "learning_rate": 0.00016199369981139433,
      "loss": 1.1016,
      "step": 4281
    },
    {
      "epoch": 0.30865710372666333,
      "grad_norm": 0.40118211172505097,
      "learning_rate": 0.00016197537874270326,
      "loss": 1.0318,
      "step": 4282
    },
    {
      "epoch": 0.3087291861890002,
      "grad_norm": 0.12114052144317598,
      "learning_rate": 0.00016195705429579958,
      "loss": 1.1425,
      "step": 4283
    },
    {
      "epoch": 0.3088012686513371,
      "grad_norm": 0.136485122353889,
      "learning_rate": 0.0001619387264716822,
      "loss": 1.1619,
      "step": 4284
    },
    {
      "epoch": 0.30887335111367403,
      "grad_norm": 0.18529368769207424,
      "learning_rate": 0.00016192039527135007,
      "loss": 1.0632,
      "step": 4285
    },
    {
      "epoch": 0.30894543357601095,
      "grad_norm": 0.1228710025314517,
      "learning_rate": 0.00016190206069580248,
      "loss": 1.0445,
      "step": 4286
    },
    {
      "epoch": 0.3090175160383479,
      "grad_norm": 0.16262058383220732,
      "learning_rate": 0.00016188372274603873,
      "loss": 1.0573,
      "step": 4287
    },
    {
      "epoch": 0.3090895985006848,
      "grad_norm": 0.25875599512530084,
      "learning_rate": 0.00016186538142305853,
      "loss": 1.1114,
      "step": 4288
    },
    {
      "epoch": 0.3091616809630217,
      "grad_norm": 0.1401565864342366,
      "learning_rate": 0.0001618470367278615,
      "loss": 1.1373,
      "step": 4289
    },
    {
      "epoch": 0.30923376342535863,
      "grad_norm": 0.1782651061268347,
      "learning_rate": 0.00016182868866144773,
      "loss": 1.072,
      "step": 4290
    },
    {
      "epoch": 0.3093058458876955,
      "grad_norm": 0.10558666763456942,
      "learning_rate": 0.0001618103372248173,
      "loss": 1.0594,
      "step": 4291
    },
    {
      "epoch": 0.3093779283500324,
      "grad_norm": 0.2432093833012497,
      "learning_rate": 0.00016179198241897047,
      "loss": 0.9988,
      "step": 4292
    },
    {
      "epoch": 0.30945001081236934,
      "grad_norm": 0.14996770725256486,
      "learning_rate": 0.0001617736242449078,
      "loss": 1.0882,
      "step": 4293
    },
    {
      "epoch": 0.30952209327470626,
      "grad_norm": 0.12317710246774367,
      "learning_rate": 0.00016175526270363,
      "loss": 1.0919,
      "step": 4294
    },
    {
      "epoch": 0.3095941757370432,
      "grad_norm": 0.13394518751010515,
      "learning_rate": 0.0001617368977961379,
      "loss": 1.1287,
      "step": 4295
    },
    {
      "epoch": 0.3096662581993801,
      "grad_norm": 0.25570599643437497,
      "learning_rate": 0.00016171852952343255,
      "loss": 1.1195,
      "step": 4296
    },
    {
      "epoch": 0.309738340661717,
      "grad_norm": 0.145645723648788,
      "learning_rate": 0.00016170015788651517,
      "loss": 1.1112,
      "step": 4297
    },
    {
      "epoch": 0.30981042312405394,
      "grad_norm": 2.4584367091603303,
      "learning_rate": 0.0001616817828863872,
      "loss": 1.1161,
      "step": 4298
    },
    {
      "epoch": 0.30988250558639086,
      "grad_norm": 0.19920420828858712,
      "learning_rate": 0.00016166340452405028,
      "loss": 1.0677,
      "step": 4299
    },
    {
      "epoch": 0.3099545880487277,
      "grad_norm": 0.12896302318837957,
      "learning_rate": 0.00016164502280050614,
      "loss": 1.1147,
      "step": 4300
    },
    {
      "epoch": 0.31002667051106464,
      "grad_norm": 0.12219415967352988,
      "learning_rate": 0.00016162663771675674,
      "loss": 1.0124,
      "step": 4301
    },
    {
      "epoch": 0.31009875297340156,
      "grad_norm": 0.13321425325050235,
      "learning_rate": 0.0001616082492738043,
      "loss": 1.1051,
      "step": 4302
    },
    {
      "epoch": 0.3101708354357385,
      "grad_norm": 0.15140797967408445,
      "learning_rate": 0.00016158985747265108,
      "loss": 1.0094,
      "step": 4303
    },
    {
      "epoch": 0.3102429178980754,
      "grad_norm": 0.13177718956797282,
      "learning_rate": 0.0001615714623142996,
      "loss": 1.1132,
      "step": 4304
    },
    {
      "epoch": 0.3103150003604123,
      "grad_norm": 0.26370208381966553,
      "learning_rate": 0.00016155306379975266,
      "loss": 1.1035,
      "step": 4305
    },
    {
      "epoch": 0.31038708282274924,
      "grad_norm": 0.14511231245042644,
      "learning_rate": 0.00016153466193001304,
      "loss": 1.0036,
      "step": 4306
    },
    {
      "epoch": 0.31045916528508616,
      "grad_norm": 0.2682994849303114,
      "learning_rate": 0.00016151625670608383,
      "loss": 1.0893,
      "step": 4307
    },
    {
      "epoch": 0.3105312477474231,
      "grad_norm": 0.17486033217081864,
      "learning_rate": 0.00016149784812896832,
      "loss": 1.1167,
      "step": 4308
    },
    {
      "epoch": 0.31060333020975994,
      "grad_norm": 0.1142469183037092,
      "learning_rate": 0.00016147943619966988,
      "loss": 1.1345,
      "step": 4309
    },
    {
      "epoch": 0.31067541267209686,
      "grad_norm": 0.13416381663698743,
      "learning_rate": 0.00016146102091919216,
      "loss": 1.0514,
      "step": 4310
    },
    {
      "epoch": 0.3107474951344338,
      "grad_norm": 0.12571417293308584,
      "learning_rate": 0.00016144260228853895,
      "loss": 1.1165,
      "step": 4311
    },
    {
      "epoch": 0.3108195775967707,
      "grad_norm": 0.1361765086720223,
      "learning_rate": 0.00016142418030871424,
      "loss": 1.0714,
      "step": 4312
    },
    {
      "epoch": 0.3108916600591076,
      "grad_norm": 0.14270609700071651,
      "learning_rate": 0.0001614057549807222,
      "loss": 1.1057,
      "step": 4313
    },
    {
      "epoch": 0.31096374252144454,
      "grad_norm": 0.3272876835474287,
      "learning_rate": 0.00016138732630556715,
      "loss": 1.0685,
      "step": 4314
    },
    {
      "epoch": 0.31103582498378146,
      "grad_norm": 0.12432500924341036,
      "learning_rate": 0.00016136889428425364,
      "loss": 1.1394,
      "step": 4315
    },
    {
      "epoch": 0.3111079074461184,
      "grad_norm": 0.9602368132792168,
      "learning_rate": 0.0001613504589177864,
      "loss": 1.0659,
      "step": 4316
    },
    {
      "epoch": 0.31117998990845525,
      "grad_norm": 0.12081996572994107,
      "learning_rate": 0.00016133202020717023,
      "loss": 1.0567,
      "step": 4317
    },
    {
      "epoch": 0.31125207237079217,
      "grad_norm": 0.10989745384450278,
      "learning_rate": 0.00016131357815341027,
      "loss": 1.0799,
      "step": 4318
    },
    {
      "epoch": 0.3113241548331291,
      "grad_norm": 0.12610517605228475,
      "learning_rate": 0.0001612951327575118,
      "loss": 1.0719,
      "step": 4319
    },
    {
      "epoch": 0.311396237295466,
      "grad_norm": 0.25484530319444876,
      "learning_rate": 0.00016127668402048025,
      "loss": 1.0091,
      "step": 4320
    },
    {
      "epoch": 0.3114683197578029,
      "grad_norm": 0.23748856552688713,
      "learning_rate": 0.0001612582319433212,
      "loss": 1.0042,
      "step": 4321
    },
    {
      "epoch": 0.31154040222013984,
      "grad_norm": 0.144455478038569,
      "learning_rate": 0.00016123977652704046,
      "loss": 0.9879,
      "step": 4322
    },
    {
      "epoch": 0.31161248468247676,
      "grad_norm": 0.3166755894175961,
      "learning_rate": 0.00016122131777264405,
      "loss": 1.0127,
      "step": 4323
    },
    {
      "epoch": 0.3116845671448137,
      "grad_norm": 0.11452277446894667,
      "learning_rate": 0.0001612028556811381,
      "loss": 1.0731,
      "step": 4324
    },
    {
      "epoch": 0.3117566496071506,
      "grad_norm": 0.1018141544365838,
      "learning_rate": 0.00016118439025352902,
      "loss": 1.0853,
      "step": 4325
    },
    {
      "epoch": 0.31182873206948747,
      "grad_norm": 0.11685316307388426,
      "learning_rate": 0.00016116592149082327,
      "loss": 1.0212,
      "step": 4326
    },
    {
      "epoch": 0.3119008145318244,
      "grad_norm": 0.13849082672372431,
      "learning_rate": 0.00016114744939402763,
      "loss": 1.0606,
      "step": 4327
    },
    {
      "epoch": 0.3119728969941613,
      "grad_norm": 0.18667707966302186,
      "learning_rate": 0.00016112897396414893,
      "loss": 1.0781,
      "step": 4328
    },
    {
      "epoch": 0.31204497945649823,
      "grad_norm": 1.6620690327006025,
      "learning_rate": 0.00016111049520219428,
      "loss": 1.0968,
      "step": 4329
    },
    {
      "epoch": 0.31211706191883515,
      "grad_norm": 0.14834906120966568,
      "learning_rate": 0.00016109201310917092,
      "loss": 1.0084,
      "step": 4330
    },
    {
      "epoch": 0.31218914438117207,
      "grad_norm": 0.20271236700846387,
      "learning_rate": 0.0001610735276860863,
      "loss": 1.0821,
      "step": 4331
    },
    {
      "epoch": 0.312261226843509,
      "grad_norm": 0.11227740579623151,
      "learning_rate": 0.00016105503893394806,
      "loss": 1.0842,
      "step": 4332
    },
    {
      "epoch": 0.3123333093058459,
      "grad_norm": 0.1429956225923441,
      "learning_rate": 0.00016103654685376397,
      "loss": 1.0453,
      "step": 4333
    },
    {
      "epoch": 0.3124053917681828,
      "grad_norm": 0.12563365098985474,
      "learning_rate": 0.00016101805144654204,
      "loss": 1.0603,
      "step": 4334
    },
    {
      "epoch": 0.3124774742305197,
      "grad_norm": 0.1246826726674022,
      "learning_rate": 0.0001609995527132904,
      "loss": 1.0982,
      "step": 4335
    },
    {
      "epoch": 0.3125495566928566,
      "grad_norm": 0.14777446146416107,
      "learning_rate": 0.00016098105065501743,
      "loss": 1.0063,
      "step": 4336
    },
    {
      "epoch": 0.31262163915519353,
      "grad_norm": 0.15751658246624667,
      "learning_rate": 0.00016096254527273165,
      "loss": 1.0282,
      "step": 4337
    },
    {
      "epoch": 0.31269372161753045,
      "grad_norm": 0.18976145013730025,
      "learning_rate": 0.00016094403656744178,
      "loss": 1.0541,
      "step": 4338
    },
    {
      "epoch": 0.31276580407986737,
      "grad_norm": 0.09698359657968338,
      "learning_rate": 0.0001609255245401567,
      "loss": 1.0091,
      "step": 4339
    },
    {
      "epoch": 0.3128378865422043,
      "grad_norm": 0.1712761348319219,
      "learning_rate": 0.00016090700919188544,
      "loss": 1.1324,
      "step": 4340
    },
    {
      "epoch": 0.3129099690045412,
      "grad_norm": 0.12656013673126393,
      "learning_rate": 0.00016088849052363732,
      "loss": 1.0521,
      "step": 4341
    },
    {
      "epoch": 0.31298205146687813,
      "grad_norm": 0.11341375812309305,
      "learning_rate": 0.0001608699685364217,
      "loss": 1.0165,
      "step": 4342
    },
    {
      "epoch": 0.313054133929215,
      "grad_norm": 0.1320386778659315,
      "learning_rate": 0.00016085144323124828,
      "loss": 1.1532,
      "step": 4343
    },
    {
      "epoch": 0.3131262163915519,
      "grad_norm": 0.39598401340149303,
      "learning_rate": 0.0001608329146091268,
      "loss": 1.0475,
      "step": 4344
    },
    {
      "epoch": 0.31319829885388883,
      "grad_norm": 0.1408016358813464,
      "learning_rate": 0.00016081438267106726,
      "loss": 1.0141,
      "step": 4345
    },
    {
      "epoch": 0.31327038131622575,
      "grad_norm": 0.4696865419719223,
      "learning_rate": 0.00016079584741807975,
      "loss": 1.1083,
      "step": 4346
    },
    {
      "epoch": 0.3133424637785627,
      "grad_norm": 0.1155175249533594,
      "learning_rate": 0.00016077730885117472,
      "loss": 1.0362,
      "step": 4347
    },
    {
      "epoch": 0.3134145462408996,
      "grad_norm": 0.13803952988131482,
      "learning_rate": 0.0001607587669713626,
      "loss": 1.1,
      "step": 4348
    },
    {
      "epoch": 0.3134866287032365,
      "grad_norm": 0.11047755934611994,
      "learning_rate": 0.00016074022177965414,
      "loss": 1.0962,
      "step": 4349
    },
    {
      "epoch": 0.31355871116557343,
      "grad_norm": 0.18554756781413537,
      "learning_rate": 0.00016072167327706017,
      "loss": 1.0598,
      "step": 4350
    },
    {
      "epoch": 0.31363079362791035,
      "grad_norm": 0.695027779895281,
      "learning_rate": 0.0001607031214645918,
      "loss": 1.1464,
      "step": 4351
    },
    {
      "epoch": 0.3137028760902472,
      "grad_norm": 0.11785986924905502,
      "learning_rate": 0.0001606845663432602,
      "loss": 1.0158,
      "step": 4352
    },
    {
      "epoch": 0.31377495855258414,
      "grad_norm": 0.10408065730065445,
      "learning_rate": 0.0001606660079140769,
      "loss": 1.0867,
      "step": 4353
    },
    {
      "epoch": 0.31384704101492106,
      "grad_norm": 0.1287074047389866,
      "learning_rate": 0.0001606474461780534,
      "loss": 1.013,
      "step": 4354
    },
    {
      "epoch": 0.313919123477258,
      "grad_norm": 0.11706279727141827,
      "learning_rate": 0.0001606288811362015,
      "loss": 1.1126,
      "step": 4355
    },
    {
      "epoch": 0.3139912059395949,
      "grad_norm": 0.6038821231646814,
      "learning_rate": 0.00016061031278953318,
      "loss": 1.018,
      "step": 4356
    },
    {
      "epoch": 0.3140632884019318,
      "grad_norm": 0.1351758120215716,
      "learning_rate": 0.0001605917411390606,
      "loss": 1.0971,
      "step": 4357
    },
    {
      "epoch": 0.31413537086426874,
      "grad_norm": 0.5724673220539627,
      "learning_rate": 0.00016057316618579606,
      "loss": 1.182,
      "step": 4358
    },
    {
      "epoch": 0.31420745332660566,
      "grad_norm": 0.1087104089058065,
      "learning_rate": 0.00016055458793075202,
      "loss": 1.1099,
      "step": 4359
    },
    {
      "epoch": 0.3142795357889426,
      "grad_norm": 0.13948372188122254,
      "learning_rate": 0.0001605360063749412,
      "loss": 1.0359,
      "step": 4360
    },
    {
      "epoch": 0.31435161825127944,
      "grad_norm": 0.11591657492619829,
      "learning_rate": 0.00016051742151937655,
      "loss": 1.0358,
      "step": 4361
    },
    {
      "epoch": 0.31442370071361636,
      "grad_norm": 0.13935148380540074,
      "learning_rate": 0.00016049883336507096,
      "loss": 1.0693,
      "step": 4362
    },
    {
      "epoch": 0.3144957831759533,
      "grad_norm": 0.1531954376422785,
      "learning_rate": 0.0001604802419130377,
      "loss": 0.9644,
      "step": 4363
    },
    {
      "epoch": 0.3145678656382902,
      "grad_norm": 0.15544192601850534,
      "learning_rate": 0.0001604616471642902,
      "loss": 1.1444,
      "step": 4364
    },
    {
      "epoch": 0.3146399481006271,
      "grad_norm": 0.24985936288516475,
      "learning_rate": 0.00016044304911984203,
      "loss": 1.1013,
      "step": 4365
    },
    {
      "epoch": 0.31471203056296404,
      "grad_norm": 0.1063715105426107,
      "learning_rate": 0.0001604244477807069,
      "loss": 1.0486,
      "step": 4366
    },
    {
      "epoch": 0.31478411302530096,
      "grad_norm": 0.11920515649646111,
      "learning_rate": 0.00016040584314789888,
      "loss": 1.0672,
      "step": 4367
    },
    {
      "epoch": 0.3148561954876379,
      "grad_norm": 0.313924991386704,
      "learning_rate": 0.00016038723522243193,
      "loss": 1.087,
      "step": 4368
    },
    {
      "epoch": 0.31492827794997474,
      "grad_norm": 0.31602638730967403,
      "learning_rate": 0.00016036862400532046,
      "loss": 1.0959,
      "step": 4369
    },
    {
      "epoch": 0.31500036041231166,
      "grad_norm": 0.17416903535212488,
      "learning_rate": 0.00016035000949757888,
      "loss": 1.0058,
      "step": 4370
    },
    {
      "epoch": 0.3150724428746486,
      "grad_norm": 0.12253078070374258,
      "learning_rate": 0.00016033139170022192,
      "loss": 1.0487,
      "step": 4371
    },
    {
      "epoch": 0.3151445253369855,
      "grad_norm": 0.11785106359199891,
      "learning_rate": 0.00016031277061426434,
      "loss": 0.9927,
      "step": 4372
    },
    {
      "epoch": 0.3152166077993224,
      "grad_norm": 0.18898386209889728,
      "learning_rate": 0.00016029414624072117,
      "loss": 1.1006,
      "step": 4373
    },
    {
      "epoch": 0.31528869026165934,
      "grad_norm": 0.15082130308165229,
      "learning_rate": 0.00016027551858060766,
      "loss": 1.0556,
      "step": 4374
    },
    {
      "epoch": 0.31536077272399626,
      "grad_norm": 0.19615219919974516,
      "learning_rate": 0.0001602568876349391,
      "loss": 1.0682,
      "step": 4375
    },
    {
      "epoch": 0.3154328551863332,
      "grad_norm": 0.18651947339659578,
      "learning_rate": 0.00016023825340473114,
      "loss": 1.0761,
      "step": 4376
    },
    {
      "epoch": 0.3155049376486701,
      "grad_norm": 0.1331182984644833,
      "learning_rate": 0.00016021961589099943,
      "loss": 1.0888,
      "step": 4377
    },
    {
      "epoch": 0.31557702011100697,
      "grad_norm": 0.11611207223069922,
      "learning_rate": 0.00016020097509475995,
      "loss": 0.9945,
      "step": 4378
    },
    {
      "epoch": 0.3156491025733439,
      "grad_norm": 0.3102712242905892,
      "learning_rate": 0.00016018233101702873,
      "loss": 1.0391,
      "step": 4379
    },
    {
      "epoch": 0.3157211850356808,
      "grad_norm": 0.12086719387544897,
      "learning_rate": 0.00016016368365882205,
      "loss": 1.061,
      "step": 4380
    },
    {
      "epoch": 0.3157932674980177,
      "grad_norm": 0.12800233765586375,
      "learning_rate": 0.00016014503302115639,
      "loss": 1.0374,
      "step": 4381
    },
    {
      "epoch": 0.31586534996035465,
      "grad_norm": 0.14173215878345571,
      "learning_rate": 0.00016012637910504837,
      "loss": 1.0941,
      "step": 4382
    },
    {
      "epoch": 0.31593743242269157,
      "grad_norm": 0.13192669732051718,
      "learning_rate": 0.00016010772191151474,
      "loss": 1.0794,
      "step": 4383
    },
    {
      "epoch": 0.3160095148850285,
      "grad_norm": 0.13504027701585197,
      "learning_rate": 0.00016008906144157256,
      "loss": 1.0494,
      "step": 4384
    },
    {
      "epoch": 0.3160815973473654,
      "grad_norm": 0.1468509348546437,
      "learning_rate": 0.00016007039769623895,
      "loss": 1.0239,
      "step": 4385
    },
    {
      "epoch": 0.3161536798097023,
      "grad_norm": 0.11898092710439492,
      "learning_rate": 0.00016005173067653127,
      "loss": 1.079,
      "step": 4386
    },
    {
      "epoch": 0.3162257622720392,
      "grad_norm": 0.2175624930854057,
      "learning_rate": 0.00016003306038346704,
      "loss": 1.082,
      "step": 4387
    },
    {
      "epoch": 0.3162978447343761,
      "grad_norm": 0.10649499508268576,
      "learning_rate": 0.0001600143868180639,
      "loss": 1.0392,
      "step": 4388
    },
    {
      "epoch": 0.31636992719671303,
      "grad_norm": 0.3331095682810077,
      "learning_rate": 0.00015999570998133985,
      "loss": 1.1154,
      "step": 4389
    },
    {
      "epoch": 0.31644200965904995,
      "grad_norm": 0.13874572764637702,
      "learning_rate": 0.0001599770298743128,
      "loss": 1.127,
      "step": 4390
    },
    {
      "epoch": 0.31651409212138687,
      "grad_norm": 0.11161262399292124,
      "learning_rate": 0.00015995834649800106,
      "loss": 1.0319,
      "step": 4391
    },
    {
      "epoch": 0.3165861745837238,
      "grad_norm": 0.6827737815023834,
      "learning_rate": 0.00015993965985342304,
      "loss": 1.1121,
      "step": 4392
    },
    {
      "epoch": 0.3166582570460607,
      "grad_norm": 0.7206560006015966,
      "learning_rate": 0.00015992096994159736,
      "loss": 1.0473,
      "step": 4393
    },
    {
      "epoch": 0.31673033950839763,
      "grad_norm": 0.13766471155351864,
      "learning_rate": 0.00015990227676354274,
      "loss": 1.0652,
      "step": 4394
    },
    {
      "epoch": 0.3168024219707345,
      "grad_norm": 0.11175931393173051,
      "learning_rate": 0.0001598835803202781,
      "loss": 0.993,
      "step": 4395
    },
    {
      "epoch": 0.3168745044330714,
      "grad_norm": 0.23306347207360897,
      "learning_rate": 0.0001598648806128226,
      "loss": 1.0271,
      "step": 4396
    },
    {
      "epoch": 0.31694658689540833,
      "grad_norm": 0.1294883670189215,
      "learning_rate": 0.00015984617764219555,
      "loss": 1.083,
      "step": 4397
    },
    {
      "epoch": 0.31701866935774525,
      "grad_norm": 0.12248399897274131,
      "learning_rate": 0.00015982747140941644,
      "loss": 1.1109,
      "step": 4398
    },
    {
      "epoch": 0.3170907518200822,
      "grad_norm": 0.11168106093812076,
      "learning_rate": 0.00015980876191550486,
      "loss": 1.0329,
      "step": 4399
    },
    {
      "epoch": 0.3171628342824191,
      "grad_norm": 0.11080767133607719,
      "learning_rate": 0.00015979004916148075,
      "loss": 0.9781,
      "step": 4400
    },
    {
      "epoch": 0.317234916744756,
      "grad_norm": 0.14292634216754893,
      "learning_rate": 0.00015977133314836402,
      "loss": 1.1376,
      "step": 4401
    },
    {
      "epoch": 0.31730699920709293,
      "grad_norm": 0.11840499022522422,
      "learning_rate": 0.00015975261387717496,
      "loss": 1.0002,
      "step": 4402
    },
    {
      "epoch": 0.31737908166942985,
      "grad_norm": 0.11027657484324242,
      "learning_rate": 0.00015973389134893383,
      "loss": 1.0718,
      "step": 4403
    },
    {
      "epoch": 0.3174511641317667,
      "grad_norm": 0.2272108613169814,
      "learning_rate": 0.00015971516556466124,
      "loss": 1.0084,
      "step": 4404
    },
    {
      "epoch": 0.31752324659410364,
      "grad_norm": 0.1132450731650152,
      "learning_rate": 0.00015969643652537788,
      "loss": 1.1012,
      "step": 4405
    },
    {
      "epoch": 0.31759532905644056,
      "grad_norm": 0.10553081339233165,
      "learning_rate": 0.00015967770423210467,
      "loss": 0.9853,
      "step": 4406
    },
    {
      "epoch": 0.3176674115187775,
      "grad_norm": 0.12286476215678765,
      "learning_rate": 0.00015965896868586272,
      "loss": 1.1023,
      "step": 4407
    },
    {
      "epoch": 0.3177394939811144,
      "grad_norm": 0.15585960884174532,
      "learning_rate": 0.00015964022988767325,
      "loss": 1.0248,
      "step": 4408
    },
    {
      "epoch": 0.3178115764434513,
      "grad_norm": 0.15618353358807116,
      "learning_rate": 0.00015962148783855767,
      "loss": 1.0502,
      "step": 4409
    },
    {
      "epoch": 0.31788365890578824,
      "grad_norm": 0.11464302966745177,
      "learning_rate": 0.00015960274253953762,
      "loss": 1.0317,
      "step": 4410
    },
    {
      "epoch": 0.31795574136812516,
      "grad_norm": 0.127821116430299,
      "learning_rate": 0.0001595839939916349,
      "loss": 1.0826,
      "step": 4411
    },
    {
      "epoch": 0.3180278238304621,
      "grad_norm": 0.1659564884416768,
      "learning_rate": 0.0001595652421958714,
      "loss": 1.0187,
      "step": 4412
    },
    {
      "epoch": 0.31809990629279894,
      "grad_norm": 0.12856613183198995,
      "learning_rate": 0.00015954648715326938,
      "loss": 1.097,
      "step": 4413
    },
    {
      "epoch": 0.31817198875513586,
      "grad_norm": 0.13540692406380653,
      "learning_rate": 0.0001595277288648511,
      "loss": 1.1157,
      "step": 4414
    },
    {
      "epoch": 0.3182440712174728,
      "grad_norm": 0.5873845571086804,
      "learning_rate": 0.000159508967331639,
      "loss": 1.1767,
      "step": 4415
    },
    {
      "epoch": 0.3183161536798097,
      "grad_norm": 0.11774694931875715,
      "learning_rate": 0.0001594902025546558,
      "loss": 1.0136,
      "step": 4416
    },
    {
      "epoch": 0.3183882361421466,
      "grad_norm": 0.12063245442415739,
      "learning_rate": 0.00015947143453492435,
      "loss": 1.1371,
      "step": 4417
    },
    {
      "epoch": 0.31846031860448354,
      "grad_norm": 0.15102715866608205,
      "learning_rate": 0.00015945266327346767,
      "loss": 1.0149,
      "step": 4418
    },
    {
      "epoch": 0.31853240106682046,
      "grad_norm": 0.1257503602585478,
      "learning_rate": 0.000159433888771309,
      "loss": 1.1157,
      "step": 4419
    },
    {
      "epoch": 0.3186044835291574,
      "grad_norm": 0.10532141938901689,
      "learning_rate": 0.00015941511102947163,
      "loss": 1.0958,
      "step": 4420
    },
    {
      "epoch": 0.31867656599149424,
      "grad_norm": 0.14422936111422172,
      "learning_rate": 0.00015939633004897918,
      "loss": 1.0609,
      "step": 4421
    },
    {
      "epoch": 0.31874864845383116,
      "grad_norm": 0.1077465338866123,
      "learning_rate": 0.00015937754583085538,
      "loss": 0.9904,
      "step": 4422
    },
    {
      "epoch": 0.3188207309161681,
      "grad_norm": 0.11972921040405847,
      "learning_rate": 0.0001593587583761241,
      "loss": 1.0927,
      "step": 4423
    },
    {
      "epoch": 0.318892813378505,
      "grad_norm": 0.12132653839282986,
      "learning_rate": 0.0001593399676858095,
      "loss": 1.093,
      "step": 4424
    },
    {
      "epoch": 0.3189648958408419,
      "grad_norm": 0.12819651127165532,
      "learning_rate": 0.00015932117376093574,
      "loss": 1.0501,
      "step": 4425
    },
    {
      "epoch": 0.31903697830317884,
      "grad_norm": 0.12142043335698299,
      "learning_rate": 0.00015930237660252735,
      "loss": 1.1236,
      "step": 4426
    },
    {
      "epoch": 0.31910906076551576,
      "grad_norm": 0.20817301949279943,
      "learning_rate": 0.00015928357621160886,
      "loss": 1.0616,
      "step": 4427
    },
    {
      "epoch": 0.3191811432278527,
      "grad_norm": 0.15596100896214246,
      "learning_rate": 0.00015926477258920515,
      "loss": 1.044,
      "step": 4428
    },
    {
      "epoch": 0.3192532256901896,
      "grad_norm": 0.14447785939810692,
      "learning_rate": 0.0001592459657363411,
      "loss": 1.1107,
      "step": 4429
    },
    {
      "epoch": 0.31932530815252647,
      "grad_norm": 0.18747400645314127,
      "learning_rate": 0.0001592271556540419,
      "loss": 1.0307,
      "step": 4430
    },
    {
      "epoch": 0.3193973906148634,
      "grad_norm": 0.12567931564406465,
      "learning_rate": 0.00015920834234333284,
      "loss": 1.0875,
      "step": 4431
    },
    {
      "epoch": 0.3194694730772003,
      "grad_norm": 0.1592040239093727,
      "learning_rate": 0.00015918952580523942,
      "loss": 1.0931,
      "step": 4432
    },
    {
      "epoch": 0.3195415555395372,
      "grad_norm": 0.29566145872422017,
      "learning_rate": 0.00015917070604078734,
      "loss": 1.1128,
      "step": 4433
    },
    {
      "epoch": 0.31961363800187415,
      "grad_norm": 0.1063451178474401,
      "learning_rate": 0.00015915188305100245,
      "loss": 1.0914,
      "step": 4434
    },
    {
      "epoch": 0.31968572046421107,
      "grad_norm": 0.1215253976602827,
      "learning_rate": 0.00015913305683691072,
      "loss": 1.1143,
      "step": 4435
    },
    {
      "epoch": 0.319757802926548,
      "grad_norm": 0.13307235347478663,
      "learning_rate": 0.00015911422739953835,
      "loss": 1.089,
      "step": 4436
    },
    {
      "epoch": 0.3198298853888849,
      "grad_norm": 0.7903570534714655,
      "learning_rate": 0.00015909539473991179,
      "loss": 1.2005,
      "step": 4437
    },
    {
      "epoch": 0.3199019678512218,
      "grad_norm": 0.4443750327323792,
      "learning_rate": 0.00015907655885905747,
      "loss": 1.1905,
      "step": 4438
    },
    {
      "epoch": 0.3199740503135587,
      "grad_norm": 0.24585045663896454,
      "learning_rate": 0.00015905771975800221,
      "loss": 1.0945,
      "step": 4439
    },
    {
      "epoch": 0.3200461327758956,
      "grad_norm": 0.11932968980130527,
      "learning_rate": 0.00015903887743777287,
      "loss": 1.0705,
      "step": 4440
    },
    {
      "epoch": 0.32011821523823253,
      "grad_norm": 0.38885915629645673,
      "learning_rate": 0.00015902003189939654,
      "loss": 1.0601,
      "step": 4441
    },
    {
      "epoch": 0.32019029770056945,
      "grad_norm": 0.12824939337582844,
      "learning_rate": 0.0001590011831439005,
      "loss": 1.0875,
      "step": 4442
    },
    {
      "epoch": 0.32026238016290637,
      "grad_norm": 0.13048393322223859,
      "learning_rate": 0.00015898233117231207,
      "loss": 1.1025,
      "step": 4443
    },
    {
      "epoch": 0.3203344626252433,
      "grad_norm": 0.2028293937743641,
      "learning_rate": 0.00015896347598565896,
      "loss": 1.0644,
      "step": 4444
    },
    {
      "epoch": 0.3204065450875802,
      "grad_norm": 0.1710257895253237,
      "learning_rate": 0.00015894461758496887,
      "loss": 1.02,
      "step": 4445
    },
    {
      "epoch": 0.3204786275499171,
      "grad_norm": 0.12739341765190854,
      "learning_rate": 0.00015892575597126985,
      "loss": 1.0089,
      "step": 4446
    },
    {
      "epoch": 0.320550710012254,
      "grad_norm": 0.18275914827711937,
      "learning_rate": 0.0001589068911455899,
      "loss": 1.0847,
      "step": 4447
    },
    {
      "epoch": 0.3206227924745909,
      "grad_norm": 0.1429177272787111,
      "learning_rate": 0.00015888802310895742,
      "loss": 1.0559,
      "step": 4448
    },
    {
      "epoch": 0.32069487493692783,
      "grad_norm": 0.13433497865120084,
      "learning_rate": 0.0001588691518624009,
      "loss": 0.9847,
      "step": 4449
    },
    {
      "epoch": 0.32076695739926475,
      "grad_norm": 0.1289575995872028,
      "learning_rate": 0.00015885027740694888,
      "loss": 1.0608,
      "step": 4450
    },
    {
      "epoch": 0.32083903986160167,
      "grad_norm": 0.14116103899809254,
      "learning_rate": 0.00015883139974363027,
      "loss": 1.003,
      "step": 4451
    },
    {
      "epoch": 0.3209111223239386,
      "grad_norm": 0.18280885511316464,
      "learning_rate": 0.00015881251887347404,
      "loss": 1.0394,
      "step": 4452
    },
    {
      "epoch": 0.3209832047862755,
      "grad_norm": 0.14114524472937412,
      "learning_rate": 0.0001587936347975094,
      "loss": 1.1028,
      "step": 4453
    },
    {
      "epoch": 0.32105528724861243,
      "grad_norm": 0.18617313783549613,
      "learning_rate": 0.00015877474751676567,
      "loss": 1.0071,
      "step": 4454
    },
    {
      "epoch": 0.32112736971094935,
      "grad_norm": 0.21489545074957514,
      "learning_rate": 0.0001587558570322724,
      "loss": 1.0644,
      "step": 4455
    },
    {
      "epoch": 0.3211994521732862,
      "grad_norm": 0.13829757442799373,
      "learning_rate": 0.00015873696334505922,
      "loss": 1.1003,
      "step": 4456
    },
    {
      "epoch": 0.32127153463562313,
      "grad_norm": 0.1128153650348434,
      "learning_rate": 0.00015871806645615614,
      "loss": 1.047,
      "step": 4457
    },
    {
      "epoch": 0.32134361709796005,
      "grad_norm": 0.9101115148808429,
      "learning_rate": 0.0001586991663665931,
      "loss": 1.1234,
      "step": 4458
    },
    {
      "epoch": 0.321415699560297,
      "grad_norm": 0.11779318104374674,
      "learning_rate": 0.00015868026307740035,
      "loss": 1.0368,
      "step": 4459
    },
    {
      "epoch": 0.3214877820226339,
      "grad_norm": 0.11785714994917504,
      "learning_rate": 0.0001586613565896083,
      "loss": 1.0454,
      "step": 4460
    },
    {
      "epoch": 0.3215598644849708,
      "grad_norm": 0.17877236451162118,
      "learning_rate": 0.0001586424469042475,
      "loss": 1.1091,
      "step": 4461
    },
    {
      "epoch": 0.32163194694730773,
      "grad_norm": 0.12119263602743824,
      "learning_rate": 0.00015862353402234876,
      "loss": 1.0799,
      "step": 4462
    },
    {
      "epoch": 0.32170402940964465,
      "grad_norm": 0.11913652006231494,
      "learning_rate": 0.00015860461794494287,
      "loss": 1.0748,
      "step": 4463
    },
    {
      "epoch": 0.3217761118719816,
      "grad_norm": 0.18792795868290227,
      "learning_rate": 0.00015858569867306107,
      "loss": 1.0474,
      "step": 4464
    },
    {
      "epoch": 0.32184819433431844,
      "grad_norm": 0.1310013094934633,
      "learning_rate": 0.00015856677620773457,
      "loss": 1.0296,
      "step": 4465
    },
    {
      "epoch": 0.32192027679665536,
      "grad_norm": 0.11671858598024873,
      "learning_rate": 0.0001585478505499948,
      "loss": 1.085,
      "step": 4466
    },
    {
      "epoch": 0.3219923592589923,
      "grad_norm": 0.13405334916141212,
      "learning_rate": 0.0001585289217008734,
      "loss": 1.0818,
      "step": 4467
    },
    {
      "epoch": 0.3220644417213292,
      "grad_norm": 0.11907598868165836,
      "learning_rate": 0.00015850998966140216,
      "loss": 1.0669,
      "step": 4468
    },
    {
      "epoch": 0.3221365241836661,
      "grad_norm": 0.11100948016879322,
      "learning_rate": 0.000158491054432613,
      "loss": 0.9945,
      "step": 4469
    },
    {
      "epoch": 0.32220860664600304,
      "grad_norm": 0.5780913749168233,
      "learning_rate": 0.00015847211601553816,
      "loss": 1.0388,
      "step": 4470
    },
    {
      "epoch": 0.32228068910833996,
      "grad_norm": 0.1346354758895666,
      "learning_rate": 0.00015845317441120984,
      "loss": 1.087,
      "step": 4471
    },
    {
      "epoch": 0.3223527715706769,
      "grad_norm": 0.20521853668344586,
      "learning_rate": 0.0001584342296206606,
      "loss": 1.0301,
      "step": 4472
    },
    {
      "epoch": 0.32242485403301374,
      "grad_norm": 0.2083370807362225,
      "learning_rate": 0.00015841528164492306,
      "loss": 1.17,
      "step": 4473
    },
    {
      "epoch": 0.32249693649535066,
      "grad_norm": 0.14223873355093042,
      "learning_rate": 0.0001583963304850301,
      "loss": 0.9993,
      "step": 4474
    },
    {
      "epoch": 0.3225690189576876,
      "grad_norm": 0.15422827587497842,
      "learning_rate": 0.00015837737614201466,
      "loss": 1.0423,
      "step": 4475
    },
    {
      "epoch": 0.3226411014200245,
      "grad_norm": 0.1322250096229391,
      "learning_rate": 0.00015835841861690998,
      "loss": 1.0716,
      "step": 4476
    },
    {
      "epoch": 0.3227131838823614,
      "grad_norm": 0.12069203140658377,
      "learning_rate": 0.00015833945791074943,
      "loss": 1.054,
      "step": 4477
    },
    {
      "epoch": 0.32278526634469834,
      "grad_norm": 0.11352470827420928,
      "learning_rate": 0.00015832049402456644,
      "loss": 1.0175,
      "step": 4478
    },
    {
      "epoch": 0.32285734880703526,
      "grad_norm": 0.17201209219999086,
      "learning_rate": 0.0001583015269593948,
      "loss": 1.0481,
      "step": 4479
    },
    {
      "epoch": 0.3229294312693722,
      "grad_norm": 0.11393449326442841,
      "learning_rate": 0.00015828255671626837,
      "loss": 1.037,
      "step": 4480
    },
    {
      "epoch": 0.3230015137317091,
      "grad_norm": 0.12981125647996666,
      "learning_rate": 0.0001582635832962212,
      "loss": 1.0401,
      "step": 4481
    },
    {
      "epoch": 0.32307359619404596,
      "grad_norm": 0.10229973301649638,
      "learning_rate": 0.0001582446067002875,
      "loss": 1.0773,
      "step": 4482
    },
    {
      "epoch": 0.3231456786563829,
      "grad_norm": 0.9556760940279838,
      "learning_rate": 0.00015822562692950165,
      "loss": 1.0422,
      "step": 4483
    },
    {
      "epoch": 0.3232177611187198,
      "grad_norm": 0.45041242224218064,
      "learning_rate": 0.00015820664398489824,
      "loss": 1.0411,
      "step": 4484
    },
    {
      "epoch": 0.3232898435810567,
      "grad_norm": 0.12028412348398705,
      "learning_rate": 0.00015818765786751197,
      "loss": 1.0674,
      "step": 4485
    },
    {
      "epoch": 0.32336192604339364,
      "grad_norm": 0.16183201333131692,
      "learning_rate": 0.0001581686685783778,
      "loss": 1.0987,
      "step": 4486
    },
    {
      "epoch": 0.32343400850573056,
      "grad_norm": 0.15440033349874416,
      "learning_rate": 0.00015814967611853083,
      "loss": 1.1146,
      "step": 4487
    },
    {
      "epoch": 0.3235060909680675,
      "grad_norm": 0.12270445863881554,
      "learning_rate": 0.00015813068048900627,
      "loss": 1.0729,
      "step": 4488
    },
    {
      "epoch": 0.3235781734304044,
      "grad_norm": 0.10575259410129238,
      "learning_rate": 0.00015811168169083955,
      "loss": 1.093,
      "step": 4489
    },
    {
      "epoch": 0.3236502558927413,
      "grad_norm": 1.5982089391816412,
      "learning_rate": 0.00015809267972506633,
      "loss": 0.9768,
      "step": 4490
    },
    {
      "epoch": 0.3237223383550782,
      "grad_norm": 0.11461040576781249,
      "learning_rate": 0.00015807367459272234,
      "loss": 1.0552,
      "step": 4491
    },
    {
      "epoch": 0.3237944208174151,
      "grad_norm": 0.10771698136950758,
      "learning_rate": 0.00015805466629484355,
      "loss": 1.0907,
      "step": 4492
    },
    {
      "epoch": 0.323866503279752,
      "grad_norm": 0.12425453612774426,
      "learning_rate": 0.00015803565483246606,
      "loss": 1.0182,
      "step": 4493
    },
    {
      "epoch": 0.32393858574208895,
      "grad_norm": 0.14197482444098986,
      "learning_rate": 0.00015801664020662618,
      "loss": 0.976,
      "step": 4494
    },
    {
      "epoch": 0.32401066820442587,
      "grad_norm": 0.12581549640762582,
      "learning_rate": 0.0001579976224183604,
      "loss": 1.0153,
      "step": 4495
    },
    {
      "epoch": 0.3240827506667628,
      "grad_norm": 0.3035690663503308,
      "learning_rate": 0.0001579786014687053,
      "loss": 1.102,
      "step": 4496
    },
    {
      "epoch": 0.3241548331290997,
      "grad_norm": 0.12089413838725141,
      "learning_rate": 0.00015795957735869774,
      "loss": 1.0703,
      "step": 4497
    },
    {
      "epoch": 0.3242269155914366,
      "grad_norm": 0.12276147485654297,
      "learning_rate": 0.0001579405500893747,
      "loss": 1.1201,
      "step": 4498
    },
    {
      "epoch": 0.3242989980537735,
      "grad_norm": 0.6071554127226881,
      "learning_rate": 0.00015792151966177335,
      "loss": 0.9885,
      "step": 4499
    },
    {
      "epoch": 0.3243710805161104,
      "grad_norm": 0.12158770492289198,
      "learning_rate": 0.00015790248607693098,
      "loss": 1.0446,
      "step": 4500
    },
    {
      "epoch": 0.32444316297844733,
      "grad_norm": 0.14712677000469196,
      "learning_rate": 0.00015788344933588512,
      "loss": 1.0577,
      "step": 4501
    },
    {
      "epoch": 0.32451524544078425,
      "grad_norm": 0.1314141413238135,
      "learning_rate": 0.0001578644094396734,
      "loss": 1.0554,
      "step": 4502
    },
    {
      "epoch": 0.32458732790312117,
      "grad_norm": 0.19560002168918628,
      "learning_rate": 0.0001578453663893337,
      "loss": 1.0328,
      "step": 4503
    },
    {
      "epoch": 0.3246594103654581,
      "grad_norm": 0.15044709367978884,
      "learning_rate": 0.00015782632018590407,
      "loss": 0.9536,
      "step": 4504
    },
    {
      "epoch": 0.324731492827795,
      "grad_norm": 0.13143225435282488,
      "learning_rate": 0.0001578072708304226,
      "loss": 1.0941,
      "step": 4505
    },
    {
      "epoch": 0.32480357529013193,
      "grad_norm": 0.37548677759882493,
      "learning_rate": 0.00015778821832392777,
      "loss": 1.1222,
      "step": 4506
    },
    {
      "epoch": 0.32487565775246885,
      "grad_norm": 0.4676297737606758,
      "learning_rate": 0.00015776916266745805,
      "loss": 1.0392,
      "step": 4507
    },
    {
      "epoch": 0.3249477402148057,
      "grad_norm": 0.1365506007813154,
      "learning_rate": 0.0001577501038620521,
      "loss": 1.0926,
      "step": 4508
    },
    {
      "epoch": 0.32501982267714263,
      "grad_norm": 0.17130444219204255,
      "learning_rate": 0.0001577310419087489,
      "loss": 1.0512,
      "step": 4509
    },
    {
      "epoch": 0.32509190513947955,
      "grad_norm": 0.228551558810494,
      "learning_rate": 0.00015771197680858738,
      "loss": 1.0015,
      "step": 4510
    },
    {
      "epoch": 0.3251639876018165,
      "grad_norm": 0.11727404784904573,
      "learning_rate": 0.00015769290856260684,
      "loss": 1.0767,
      "step": 4511
    },
    {
      "epoch": 0.3252360700641534,
      "grad_norm": 0.16675741076026201,
      "learning_rate": 0.00015767383717184668,
      "loss": 1.1243,
      "step": 4512
    },
    {
      "epoch": 0.3253081525264903,
      "grad_norm": 0.12204443974078284,
      "learning_rate": 0.0001576547626373464,
      "loss": 1.1316,
      "step": 4513
    },
    {
      "epoch": 0.32538023498882723,
      "grad_norm": 0.1045282157578462,
      "learning_rate": 0.00015763568496014575,
      "loss": 1.0177,
      "step": 4514
    },
    {
      "epoch": 0.32545231745116415,
      "grad_norm": 0.18350349275534045,
      "learning_rate": 0.00015761660414128468,
      "loss": 1.0699,
      "step": 4515
    },
    {
      "epoch": 0.3255243999135011,
      "grad_norm": 0.12057289281064408,
      "learning_rate": 0.0001575975201818032,
      "loss": 1.0252,
      "step": 4516
    },
    {
      "epoch": 0.32559648237583794,
      "grad_norm": 0.11897112235981494,
      "learning_rate": 0.00015757843308274164,
      "loss": 1.0385,
      "step": 4517
    },
    {
      "epoch": 0.32566856483817486,
      "grad_norm": 0.15121246643262204,
      "learning_rate": 0.00015755934284514029,
      "loss": 1.0119,
      "step": 4518
    },
    {
      "epoch": 0.3257406473005118,
      "grad_norm": 0.20214208596862973,
      "learning_rate": 0.00015754024947003985,
      "loss": 1.1274,
      "step": 4519
    },
    {
      "epoch": 0.3258127297628487,
      "grad_norm": 0.17500212648823832,
      "learning_rate": 0.00015752115295848104,
      "loss": 0.9692,
      "step": 4520
    },
    {
      "epoch": 0.3258848122251856,
      "grad_norm": 0.24050020994189295,
      "learning_rate": 0.0001575020533115048,
      "loss": 1.0576,
      "step": 4521
    },
    {
      "epoch": 0.32595689468752254,
      "grad_norm": 0.1427987470640518,
      "learning_rate": 0.00015748295053015222,
      "loss": 1.0742,
      "step": 4522
    },
    {
      "epoch": 0.32602897714985946,
      "grad_norm": 0.159291546755217,
      "learning_rate": 0.00015746384461546458,
      "loss": 1.0866,
      "step": 4523
    },
    {
      "epoch": 0.3261010596121964,
      "grad_norm": 0.11472963795498203,
      "learning_rate": 0.0001574447355684834,
      "loss": 1.0673,
      "step": 4524
    },
    {
      "epoch": 0.32617314207453324,
      "grad_norm": 0.10974338823300839,
      "learning_rate": 0.0001574256233902501,
      "loss": 1.0546,
      "step": 4525
    },
    {
      "epoch": 0.32624522453687016,
      "grad_norm": 0.14649123553613252,
      "learning_rate": 0.00015740650808180662,
      "loss": 0.9628,
      "step": 4526
    },
    {
      "epoch": 0.3263173069992071,
      "grad_norm": 0.15926789292824947,
      "learning_rate": 0.00015738738964419488,
      "loss": 0.9951,
      "step": 4527
    },
    {
      "epoch": 0.326389389461544,
      "grad_norm": 0.19187358908364274,
      "learning_rate": 0.00015736826807845702,
      "loss": 1.0154,
      "step": 4528
    },
    {
      "epoch": 0.3264614719238809,
      "grad_norm": 0.11761097617341888,
      "learning_rate": 0.0001573491433856353,
      "loss": 1.0327,
      "step": 4529
    },
    {
      "epoch": 0.32653355438621784,
      "grad_norm": 0.12012776896013383,
      "learning_rate": 0.0001573300155667722,
      "loss": 1.0183,
      "step": 4530
    },
    {
      "epoch": 0.32660563684855476,
      "grad_norm": 0.18271582719778445,
      "learning_rate": 0.00015731088462291038,
      "loss": 1.0221,
      "step": 4531
    },
    {
      "epoch": 0.3266777193108917,
      "grad_norm": 2.693157124757424,
      "learning_rate": 0.00015729175055509263,
      "loss": 1.1567,
      "step": 4532
    },
    {
      "epoch": 0.3267498017732286,
      "grad_norm": 0.12228743943077944,
      "learning_rate": 0.0001572726133643619,
      "loss": 1.0471,
      "step": 4533
    },
    {
      "epoch": 0.32682188423556546,
      "grad_norm": 0.12133520021950263,
      "learning_rate": 0.00015725347305176138,
      "loss": 1.0345,
      "step": 4534
    },
    {
      "epoch": 0.3268939666979024,
      "grad_norm": 0.1211715532564075,
      "learning_rate": 0.0001572343296183344,
      "loss": 1.058,
      "step": 4535
    },
    {
      "epoch": 0.3269660491602393,
      "grad_norm": 0.15529909851251983,
      "learning_rate": 0.00015721518306512438,
      "loss": 1.0181,
      "step": 4536
    },
    {
      "epoch": 0.3270381316225762,
      "grad_norm": 0.16287459469687407,
      "learning_rate": 0.00015719603339317507,
      "loss": 0.9994,
      "step": 4537
    },
    {
      "epoch": 0.32711021408491314,
      "grad_norm": 0.1764354927742462,
      "learning_rate": 0.00015717688060353023,
      "loss": 1.0239,
      "step": 4538
    },
    {
      "epoch": 0.32718229654725006,
      "grad_norm": 0.1653899251515726,
      "learning_rate": 0.00015715772469723386,
      "loss": 1.0781,
      "step": 4539
    },
    {
      "epoch": 0.327254379009587,
      "grad_norm": 0.18694499744389095,
      "learning_rate": 0.00015713856567533014,
      "loss": 1.1,
      "step": 4540
    },
    {
      "epoch": 0.3273264614719239,
      "grad_norm": 0.1270966731330697,
      "learning_rate": 0.00015711940353886346,
      "loss": 1.0669,
      "step": 4541
    },
    {
      "epoch": 0.3273985439342608,
      "grad_norm": 0.21782446875829484,
      "learning_rate": 0.00015710023828887824,
      "loss": 1.0415,
      "step": 4542
    },
    {
      "epoch": 0.3274706263965977,
      "grad_norm": 0.8648854423873328,
      "learning_rate": 0.00015708106992641922,
      "loss": 1.071,
      "step": 4543
    },
    {
      "epoch": 0.3275427088589346,
      "grad_norm": 0.13310470031091068,
      "learning_rate": 0.0001570618984525312,
      "loss": 1.0841,
      "step": 4544
    },
    {
      "epoch": 0.3276147913212715,
      "grad_norm": 1.2349291181561053,
      "learning_rate": 0.00015704272386825926,
      "loss": 1.024,
      "step": 4545
    },
    {
      "epoch": 0.32768687378360845,
      "grad_norm": 0.19902911747418328,
      "learning_rate": 0.00015702354617464854,
      "loss": 1.055,
      "step": 4546
    },
    {
      "epoch": 0.32775895624594537,
      "grad_norm": 0.9376270038090299,
      "learning_rate": 0.00015700436537274438,
      "loss": 1.0862,
      "step": 4547
    },
    {
      "epoch": 0.3278310387082823,
      "grad_norm": 0.19334098011449208,
      "learning_rate": 0.0001569851814635924,
      "loss": 1.067,
      "step": 4548
    },
    {
      "epoch": 0.3279031211706192,
      "grad_norm": 0.2634935226109979,
      "learning_rate": 0.00015696599444823815,
      "loss": 1.1264,
      "step": 4549
    },
    {
      "epoch": 0.3279752036329561,
      "grad_norm": 0.22950708217446117,
      "learning_rate": 0.00015694680432772762,
      "loss": 1.1165,
      "step": 4550
    },
    {
      "epoch": 0.328047286095293,
      "grad_norm": 0.13000476704642935,
      "learning_rate": 0.00015692761110310674,
      "loss": 1.1266,
      "step": 4551
    },
    {
      "epoch": 0.3281193685576299,
      "grad_norm": 0.1277623794852272,
      "learning_rate": 0.0001569084147754218,
      "loss": 1.1133,
      "step": 4552
    },
    {
      "epoch": 0.32819145101996683,
      "grad_norm": 0.11053660792353333,
      "learning_rate": 0.00015688921534571913,
      "loss": 1.0993,
      "step": 4553
    },
    {
      "epoch": 0.32826353348230375,
      "grad_norm": 0.6933928558081542,
      "learning_rate": 0.0001568700128150453,
      "loss": 1.0482,
      "step": 4554
    },
    {
      "epoch": 0.32833561594464067,
      "grad_norm": 0.19297286377402717,
      "learning_rate": 0.00015685080718444699,
      "loss": 1.039,
      "step": 4555
    },
    {
      "epoch": 0.3284076984069776,
      "grad_norm": 0.14448411451172236,
      "learning_rate": 0.0001568315984549711,
      "loss": 1.0938,
      "step": 4556
    },
    {
      "epoch": 0.3284797808693145,
      "grad_norm": 0.1251899290940972,
      "learning_rate": 0.00015681238662766463,
      "loss": 1.0858,
      "step": 4557
    },
    {
      "epoch": 0.32855186333165143,
      "grad_norm": 0.24450137598866878,
      "learning_rate": 0.00015679317170357485,
      "loss": 1.1229,
      "step": 4558
    },
    {
      "epoch": 0.32862394579398835,
      "grad_norm": 0.14359721853061974,
      "learning_rate": 0.00015677395368374913,
      "loss": 1.0637,
      "step": 4559
    },
    {
      "epoch": 0.3286960282563252,
      "grad_norm": 0.11080862571603269,
      "learning_rate": 0.00015675473256923498,
      "loss": 1.0695,
      "step": 4560
    },
    {
      "epoch": 0.32876811071866213,
      "grad_norm": 0.15660201998554024,
      "learning_rate": 0.0001567355083610802,
      "loss": 1.0932,
      "step": 4561
    },
    {
      "epoch": 0.32884019318099905,
      "grad_norm": 0.13824955190053909,
      "learning_rate": 0.00015671628106033264,
      "loss": 1.0376,
      "step": 4562
    },
    {
      "epoch": 0.32891227564333597,
      "grad_norm": 0.18235775147629021,
      "learning_rate": 0.00015669705066804037,
      "loss": 1.0714,
      "step": 4563
    },
    {
      "epoch": 0.3289843581056729,
      "grad_norm": 0.12178316043472939,
      "learning_rate": 0.00015667781718525157,
      "loss": 1.0225,
      "step": 4564
    },
    {
      "epoch": 0.3290564405680098,
      "grad_norm": 0.2270166356453858,
      "learning_rate": 0.00015665858061301476,
      "loss": 1.0133,
      "step": 4565
    },
    {
      "epoch": 0.32912852303034673,
      "grad_norm": 0.13341263036404172,
      "learning_rate": 0.00015663934095237837,
      "loss": 0.9431,
      "step": 4566
    },
    {
      "epoch": 0.32920060549268365,
      "grad_norm": 0.13078130413366434,
      "learning_rate": 0.00015662009820439118,
      "loss": 1.0492,
      "step": 4567
    },
    {
      "epoch": 0.32927268795502057,
      "grad_norm": 0.11554881841082593,
      "learning_rate": 0.00015660085237010214,
      "loss": 1.0018,
      "step": 4568
    },
    {
      "epoch": 0.32934477041735744,
      "grad_norm": 0.12058684432375352,
      "learning_rate": 0.00015658160345056023,
      "loss": 1.0373,
      "step": 4569
    },
    {
      "epoch": 0.32941685287969436,
      "grad_norm": 0.1536548996463659,
      "learning_rate": 0.00015656235144681478,
      "loss": 1.0816,
      "step": 4570
    },
    {
      "epoch": 0.3294889353420313,
      "grad_norm": 0.1315631665647027,
      "learning_rate": 0.00015654309635991512,
      "loss": 1.0577,
      "step": 4571
    },
    {
      "epoch": 0.3295610178043682,
      "grad_norm": 0.1432031527763279,
      "learning_rate": 0.00015652383819091088,
      "loss": 1.0413,
      "step": 4572
    },
    {
      "epoch": 0.3296331002667051,
      "grad_norm": 0.13242874596574886,
      "learning_rate": 0.0001565045769408518,
      "loss": 1.0287,
      "step": 4573
    },
    {
      "epoch": 0.32970518272904203,
      "grad_norm": 0.11899865612103878,
      "learning_rate": 0.00015648531261078775,
      "loss": 0.9819,
      "step": 4574
    },
    {
      "epoch": 0.32977726519137895,
      "grad_norm": 0.1569757110238892,
      "learning_rate": 0.00015646604520176883,
      "loss": 1.0024,
      "step": 4575
    },
    {
      "epoch": 0.3298493476537159,
      "grad_norm": 0.1652008206259976,
      "learning_rate": 0.00015644677471484528,
      "loss": 1.0327,
      "step": 4576
    },
    {
      "epoch": 0.32992143011605274,
      "grad_norm": 0.11491039129460048,
      "learning_rate": 0.00015642750115106753,
      "loss": 1.0287,
      "step": 4577
    },
    {
      "epoch": 0.32999351257838966,
      "grad_norm": 0.3678466702856971,
      "learning_rate": 0.00015640822451148614,
      "loss": 1.0779,
      "step": 4578
    },
    {
      "epoch": 0.3300655950407266,
      "grad_norm": 0.134678827522952,
      "learning_rate": 0.0001563889447971519,
      "loss": 1.0726,
      "step": 4579
    },
    {
      "epoch": 0.3301376775030635,
      "grad_norm": 0.1677949003466692,
      "learning_rate": 0.0001563696620091157,
      "loss": 1.0114,
      "step": 4580
    },
    {
      "epoch": 0.3302097599654004,
      "grad_norm": 0.1822827084904927,
      "learning_rate": 0.00015635037614842857,
      "loss": 1.0956,
      "step": 4581
    },
    {
      "epoch": 0.33028184242773734,
      "grad_norm": 0.15521378461522312,
      "learning_rate": 0.00015633108721614187,
      "loss": 1.038,
      "step": 4582
    },
    {
      "epoch": 0.33035392489007426,
      "grad_norm": 0.38021365236323795,
      "learning_rate": 0.00015631179521330695,
      "loss": 1.0964,
      "step": 4583
    },
    {
      "epoch": 0.3304260073524112,
      "grad_norm": 0.13853801823542342,
      "learning_rate": 0.00015629250014097542,
      "loss": 1.056,
      "step": 4584
    },
    {
      "epoch": 0.3304980898147481,
      "grad_norm": 0.23101731275622128,
      "learning_rate": 0.000156273202000199,
      "loss": 1.0101,
      "step": 4585
    },
    {
      "epoch": 0.33057017227708496,
      "grad_norm": 0.15615037437176402,
      "learning_rate": 0.00015625390079202961,
      "loss": 1.0413,
      "step": 4586
    },
    {
      "epoch": 0.3306422547394219,
      "grad_norm": 0.14060110180090227,
      "learning_rate": 0.0001562345965175194,
      "loss": 1.0467,
      "step": 4587
    },
    {
      "epoch": 0.3307143372017588,
      "grad_norm": 0.13286535577617714,
      "learning_rate": 0.0001562152891777206,
      "loss": 1.1454,
      "step": 4588
    },
    {
      "epoch": 0.3307864196640957,
      "grad_norm": 0.11438121807076192,
      "learning_rate": 0.00015619597877368564,
      "loss": 1.1118,
      "step": 4589
    },
    {
      "epoch": 0.33085850212643264,
      "grad_norm": 0.2126708608769687,
      "learning_rate": 0.0001561766653064671,
      "loss": 1.1169,
      "step": 4590
    },
    {
      "epoch": 0.33093058458876956,
      "grad_norm": 0.12901428796887113,
      "learning_rate": 0.00015615734877711768,
      "loss": 0.9914,
      "step": 4591
    },
    {
      "epoch": 0.3310026670511065,
      "grad_norm": 0.2105765506555687,
      "learning_rate": 0.0001561380291866904,
      "loss": 1.0883,
      "step": 4592
    },
    {
      "epoch": 0.3310747495134434,
      "grad_norm": 0.10850417671343011,
      "learning_rate": 0.00015611870653623825,
      "loss": 1.0783,
      "step": 4593
    },
    {
      "epoch": 0.3311468319757803,
      "grad_norm": 1.1715164273359169,
      "learning_rate": 0.00015609938082681457,
      "loss": 1.1083,
      "step": 4594
    },
    {
      "epoch": 0.3312189144381172,
      "grad_norm": 0.2579027701096876,
      "learning_rate": 0.00015608005205947274,
      "loss": 1.0197,
      "step": 4595
    },
    {
      "epoch": 0.3312909969004541,
      "grad_norm": 0.11615036314147907,
      "learning_rate": 0.0001560607202352664,
      "loss": 1.0842,
      "step": 4596
    },
    {
      "epoch": 0.331363079362791,
      "grad_norm": 0.1303837708619594,
      "learning_rate": 0.00015604138535524925,
      "loss": 1.0681,
      "step": 4597
    },
    {
      "epoch": 0.33143516182512794,
      "grad_norm": 0.11643651434924221,
      "learning_rate": 0.00015602204742047525,
      "loss": 1.0705,
      "step": 4598
    },
    {
      "epoch": 0.33150724428746486,
      "grad_norm": 0.1336219003763932,
      "learning_rate": 0.00015600270643199846,
      "loss": 1.0943,
      "step": 4599
    },
    {
      "epoch": 0.3315793267498018,
      "grad_norm": 0.17241486584914176,
      "learning_rate": 0.00015598336239087317,
      "loss": 1.0623,
      "step": 4600
    },
    {
      "epoch": 0.3316514092121387,
      "grad_norm": 0.15145671925471393,
      "learning_rate": 0.0001559640152981538,
      "loss": 1.0793,
      "step": 4601
    },
    {
      "epoch": 0.3317234916744756,
      "grad_norm": 0.16495366761703842,
      "learning_rate": 0.0001559446651548949,
      "loss": 1.0869,
      "step": 4602
    },
    {
      "epoch": 0.3317955741368125,
      "grad_norm": 0.16476574189577167,
      "learning_rate": 0.00015592531196215127,
      "loss": 1.0581,
      "step": 4603
    },
    {
      "epoch": 0.3318676565991494,
      "grad_norm": 0.10832524426855236,
      "learning_rate": 0.0001559059557209778,
      "loss": 1.074,
      "step": 4604
    },
    {
      "epoch": 0.3319397390614863,
      "grad_norm": 0.1334203095535034,
      "learning_rate": 0.0001558865964324296,
      "loss": 1.075,
      "step": 4605
    },
    {
      "epoch": 0.33201182152382325,
      "grad_norm": 0.11085082891764306,
      "learning_rate": 0.00015586723409756194,
      "loss": 1.0783,
      "step": 4606
    },
    {
      "epoch": 0.33208390398616017,
      "grad_norm": 0.22090097519930776,
      "learning_rate": 0.0001558478687174302,
      "loss": 1.0498,
      "step": 4607
    },
    {
      "epoch": 0.3321559864484971,
      "grad_norm": 0.26299615875513016,
      "learning_rate": 0.00015582850029308997,
      "loss": 1.098,
      "step": 4608
    },
    {
      "epoch": 0.332228068910834,
      "grad_norm": 0.11359066877621586,
      "learning_rate": 0.00015580912882559705,
      "loss": 1.1151,
      "step": 4609
    },
    {
      "epoch": 0.3323001513731709,
      "grad_norm": 0.12909729882247878,
      "learning_rate": 0.00015578975431600732,
      "loss": 1.1142,
      "step": 4610
    },
    {
      "epoch": 0.33237223383550785,
      "grad_norm": 0.1289608122571026,
      "learning_rate": 0.00015577037676537688,
      "loss": 1.0488,
      "step": 4611
    },
    {
      "epoch": 0.3324443162978447,
      "grad_norm": 0.4113930879265359,
      "learning_rate": 0.00015575099617476195,
      "loss": 1.0375,
      "step": 4612
    },
    {
      "epoch": 0.33251639876018163,
      "grad_norm": 0.12838827797007327,
      "learning_rate": 0.00015573161254521896,
      "loss": 1.1834,
      "step": 4613
    },
    {
      "epoch": 0.33258848122251855,
      "grad_norm": 0.11584365089703856,
      "learning_rate": 0.00015571222587780451,
      "loss": 1.0366,
      "step": 4614
    },
    {
      "epoch": 0.33266056368485547,
      "grad_norm": 0.11148719603468799,
      "learning_rate": 0.00015569283617357534,
      "loss": 1.0016,
      "step": 4615
    },
    {
      "epoch": 0.3327326461471924,
      "grad_norm": 0.11679188212207786,
      "learning_rate": 0.00015567344343358835,
      "loss": 1.0134,
      "step": 4616
    },
    {
      "epoch": 0.3328047286095293,
      "grad_norm": 1.3362877913944453,
      "learning_rate": 0.00015565404765890061,
      "loss": 1.0414,
      "step": 4617
    },
    {
      "epoch": 0.33287681107186623,
      "grad_norm": 0.1499357717681979,
      "learning_rate": 0.0001556346488505694,
      "loss": 1.0809,
      "step": 4618
    },
    {
      "epoch": 0.33294889353420315,
      "grad_norm": 0.2286152454583993,
      "learning_rate": 0.00015561524700965208,
      "loss": 1.0867,
      "step": 4619
    },
    {
      "epoch": 0.33302097599654007,
      "grad_norm": 0.11097785594460748,
      "learning_rate": 0.00015559584213720626,
      "loss": 1.084,
      "step": 4620
    },
    {
      "epoch": 0.33309305845887693,
      "grad_norm": 0.34818543264853064,
      "learning_rate": 0.00015557643423428968,
      "loss": 1.2043,
      "step": 4621
    },
    {
      "epoch": 0.33316514092121385,
      "grad_norm": 0.14657122846852094,
      "learning_rate": 0.00015555702330196023,
      "loss": 1.1124,
      "step": 4622
    },
    {
      "epoch": 0.3332372233835508,
      "grad_norm": 0.10642336576710414,
      "learning_rate": 0.000155537609341276,
      "loss": 1.0699,
      "step": 4623
    },
    {
      "epoch": 0.3333093058458877,
      "grad_norm": 0.12274060246592307,
      "learning_rate": 0.00015551819235329515,
      "loss": 1.0437,
      "step": 4624
    },
    {
      "epoch": 0.3333813883082246,
      "grad_norm": 0.20593052833521583,
      "learning_rate": 0.00015549877233907622,
      "loss": 1.0604,
      "step": 4625
    },
    {
      "epoch": 0.33345347077056153,
      "grad_norm": 0.10853256303099859,
      "learning_rate": 0.0001554793492996776,
      "loss": 1.0145,
      "step": 4626
    },
    {
      "epoch": 0.33352555323289845,
      "grad_norm": 0.11725671340739854,
      "learning_rate": 0.00015545992323615817,
      "loss": 1.1256,
      "step": 4627
    },
    {
      "epoch": 0.3335976356952354,
      "grad_norm": 2.6411521362353803,
      "learning_rate": 0.00015544049414957674,
      "loss": 1.0406,
      "step": 4628
    },
    {
      "epoch": 0.33366971815757224,
      "grad_norm": 0.13304924184986586,
      "learning_rate": 0.00015542106204099242,
      "loss": 1.1329,
      "step": 4629
    },
    {
      "epoch": 0.33374180061990916,
      "grad_norm": 0.1107951209745154,
      "learning_rate": 0.0001554016269114644,
      "loss": 1.0211,
      "step": 4630
    },
    {
      "epoch": 0.3338138830822461,
      "grad_norm": 0.13713631983062557,
      "learning_rate": 0.00015538218876205205,
      "loss": 1.0646,
      "step": 4631
    },
    {
      "epoch": 0.333885965544583,
      "grad_norm": 0.19712837878293307,
      "learning_rate": 0.00015536274759381496,
      "loss": 1.0945,
      "step": 4632
    },
    {
      "epoch": 0.3339580480069199,
      "grad_norm": 0.10237917556669507,
      "learning_rate": 0.00015534330340781285,
      "loss": 1.1081,
      "step": 4633
    },
    {
      "epoch": 0.33403013046925684,
      "grad_norm": 0.41279345982466714,
      "learning_rate": 0.00015532385620510558,
      "loss": 1.0564,
      "step": 4634
    },
    {
      "epoch": 0.33410221293159376,
      "grad_norm": 3.17126321835961,
      "learning_rate": 0.0001553044059867532,
      "loss": 1.0952,
      "step": 4635
    },
    {
      "epoch": 0.3341742953939307,
      "grad_norm": 0.13047878699843676,
      "learning_rate": 0.00015528495275381597,
      "loss": 1.1046,
      "step": 4636
    },
    {
      "epoch": 0.3342463778562676,
      "grad_norm": 0.23671931136611063,
      "learning_rate": 0.00015526549650735416,
      "loss": 1.0452,
      "step": 4637
    },
    {
      "epoch": 0.33431846031860446,
      "grad_norm": 0.24917641714681396,
      "learning_rate": 0.00015524603724842843,
      "loss": 1.0551,
      "step": 4638
    },
    {
      "epoch": 0.3343905427809414,
      "grad_norm": 0.10293665766553625,
      "learning_rate": 0.00015522657497809935,
      "loss": 1.1406,
      "step": 4639
    },
    {
      "epoch": 0.3344626252432783,
      "grad_norm": 0.17339021349051462,
      "learning_rate": 0.00015520710969742793,
      "loss": 1.106,
      "step": 4640
    },
    {
      "epoch": 0.3345347077056152,
      "grad_norm": 0.1548974890642215,
      "learning_rate": 0.0001551876414074751,
      "loss": 1.0285,
      "step": 4641
    },
    {
      "epoch": 0.33460679016795214,
      "grad_norm": 1.5381075926079895,
      "learning_rate": 0.00015516817010930212,
      "loss": 1.0257,
      "step": 4642
    },
    {
      "epoch": 0.33467887263028906,
      "grad_norm": 0.12893935896980224,
      "learning_rate": 0.0001551486958039703,
      "loss": 0.9726,
      "step": 4643
    },
    {
      "epoch": 0.334750955092626,
      "grad_norm": 0.11126569810168735,
      "learning_rate": 0.0001551292184925412,
      "loss": 1.0257,
      "step": 4644
    },
    {
      "epoch": 0.3348230375549629,
      "grad_norm": 0.11583290308164416,
      "learning_rate": 0.00015510973817607648,
      "loss": 1.0735,
      "step": 4645
    },
    {
      "epoch": 0.33489512001729976,
      "grad_norm": 0.22219320204019596,
      "learning_rate": 0.00015509025485563798,
      "loss": 1.1288,
      "step": 4646
    },
    {
      "epoch": 0.3349672024796367,
      "grad_norm": 1.343158620589945,
      "learning_rate": 0.00015507076853228777,
      "loss": 1.0412,
      "step": 4647
    },
    {
      "epoch": 0.3350392849419736,
      "grad_norm": 0.18768701654416542,
      "learning_rate": 0.00015505127920708797,
      "loss": 1.1207,
      "step": 4648
    },
    {
      "epoch": 0.3351113674043105,
      "grad_norm": 0.12607672004780385,
      "learning_rate": 0.000155031786881101,
      "loss": 0.9905,
      "step": 4649
    },
    {
      "epoch": 0.33518344986664744,
      "grad_norm": 0.24494401093180834,
      "learning_rate": 0.00015501229155538926,
      "loss": 1.1005,
      "step": 4650
    },
    {
      "epoch": 0.33525553232898436,
      "grad_norm": 0.12742133092157396,
      "learning_rate": 0.0001549927932310155,
      "loss": 1.063,
      "step": 4651
    },
    {
      "epoch": 0.3353276147913213,
      "grad_norm": 0.11820198666341791,
      "learning_rate": 0.00015497329190904252,
      "loss": 1.084,
      "step": 4652
    },
    {
      "epoch": 0.3353996972536582,
      "grad_norm": 0.7283177395138121,
      "learning_rate": 0.00015495378759053336,
      "loss": 1.0751,
      "step": 4653
    },
    {
      "epoch": 0.3354717797159951,
      "grad_norm": 0.1742789399293052,
      "learning_rate": 0.0001549342802765511,
      "loss": 1.0295,
      "step": 4654
    },
    {
      "epoch": 0.335543862178332,
      "grad_norm": 0.15130518702920254,
      "learning_rate": 0.00015491476996815914,
      "loss": 0.9531,
      "step": 4655
    },
    {
      "epoch": 0.3356159446406689,
      "grad_norm": 0.1124509593577456,
      "learning_rate": 0.00015489525666642092,
      "loss": 1.0073,
      "step": 4656
    },
    {
      "epoch": 0.3356880271030058,
      "grad_norm": 0.13599012829536924,
      "learning_rate": 0.00015487574037240012,
      "loss": 1.1522,
      "step": 4657
    },
    {
      "epoch": 0.33576010956534275,
      "grad_norm": 0.1341448425633751,
      "learning_rate": 0.00015485622108716052,
      "loss": 1.0977,
      "step": 4658
    },
    {
      "epoch": 0.33583219202767967,
      "grad_norm": 0.12114882936500244,
      "learning_rate": 0.00015483669881176612,
      "loss": 1.0136,
      "step": 4659
    },
    {
      "epoch": 0.3359042744900166,
      "grad_norm": 0.26134896461803864,
      "learning_rate": 0.00015481717354728106,
      "loss": 1.0588,
      "step": 4660
    },
    {
      "epoch": 0.3359763569523535,
      "grad_norm": 0.14458983777107587,
      "learning_rate": 0.0001547976452947696,
      "loss": 1.1472,
      "step": 4661
    },
    {
      "epoch": 0.3360484394146904,
      "grad_norm": 0.1785504189030975,
      "learning_rate": 0.00015477811405529627,
      "loss": 1.0049,
      "step": 4662
    },
    {
      "epoch": 0.33612052187702735,
      "grad_norm": 0.11678922815541692,
      "learning_rate": 0.00015475857982992568,
      "loss": 1.1087,
      "step": 4663
    },
    {
      "epoch": 0.3361926043393642,
      "grad_norm": 0.19514646751237966,
      "learning_rate": 0.00015473904261972257,
      "loss": 1.0494,
      "step": 4664
    },
    {
      "epoch": 0.33626468680170113,
      "grad_norm": 0.7045017179113089,
      "learning_rate": 0.00015471950242575196,
      "loss": 1.0707,
      "step": 4665
    },
    {
      "epoch": 0.33633676926403805,
      "grad_norm": 0.1350574420731982,
      "learning_rate": 0.0001546999592490789,
      "loss": 1.0254,
      "step": 4666
    },
    {
      "epoch": 0.33640885172637497,
      "grad_norm": 0.11088520519039226,
      "learning_rate": 0.00015468041309076873,
      "loss": 1.1155,
      "step": 4667
    },
    {
      "epoch": 0.3364809341887119,
      "grad_norm": 0.13768214829133138,
      "learning_rate": 0.00015466086395188682,
      "loss": 1.0415,
      "step": 4668
    },
    {
      "epoch": 0.3365530166510488,
      "grad_norm": 0.2785737334808299,
      "learning_rate": 0.00015464131183349887,
      "loss": 1.0628,
      "step": 4669
    },
    {
      "epoch": 0.33662509911338573,
      "grad_norm": 0.7014554056004816,
      "learning_rate": 0.00015462175673667053,
      "loss": 1.1124,
      "step": 4670
    },
    {
      "epoch": 0.33669718157572265,
      "grad_norm": 0.2563292907797587,
      "learning_rate": 0.00015460219866246785,
      "loss": 0.994,
      "step": 4671
    },
    {
      "epoch": 0.3367692640380595,
      "grad_norm": 0.1295477401358837,
      "learning_rate": 0.00015458263761195678,
      "loss": 1.1009,
      "step": 4672
    },
    {
      "epoch": 0.33684134650039643,
      "grad_norm": 0.11455397766586178,
      "learning_rate": 0.00015456307358620373,
      "loss": 1.0507,
      "step": 4673
    },
    {
      "epoch": 0.33691342896273335,
      "grad_norm": 0.13310960909387495,
      "learning_rate": 0.00015454350658627498,
      "loss": 1.0696,
      "step": 4674
    },
    {
      "epoch": 0.33698551142507027,
      "grad_norm": 0.11757952098581718,
      "learning_rate": 0.00015452393661323722,
      "loss": 1.0617,
      "step": 4675
    },
    {
      "epoch": 0.3370575938874072,
      "grad_norm": 0.13191041493745273,
      "learning_rate": 0.00015450436366815705,
      "loss": 1.0446,
      "step": 4676
    },
    {
      "epoch": 0.3371296763497441,
      "grad_norm": 0.13367949168148963,
      "learning_rate": 0.00015448478775210149,
      "loss": 1.0755,
      "step": 4677
    },
    {
      "epoch": 0.33720175881208103,
      "grad_norm": 0.11131651192443776,
      "learning_rate": 0.00015446520886613758,
      "loss": 1.0728,
      "step": 4678
    },
    {
      "epoch": 0.33727384127441795,
      "grad_norm": 0.10979825397106055,
      "learning_rate": 0.0001544456270113325,
      "loss": 1.048,
      "step": 4679
    },
    {
      "epoch": 0.33734592373675487,
      "grad_norm": 0.7161267314035449,
      "learning_rate": 0.0001544260421887537,
      "loss": 1.045,
      "step": 4680
    },
    {
      "epoch": 0.33741800619909174,
      "grad_norm": 0.13543034851865576,
      "learning_rate": 0.00015440645439946863,
      "loss": 1.1409,
      "step": 4681
    },
    {
      "epoch": 0.33749008866142866,
      "grad_norm": 0.1268701754525834,
      "learning_rate": 0.00015438686364454513,
      "loss": 1.0634,
      "step": 4682
    },
    {
      "epoch": 0.3375621711237656,
      "grad_norm": 0.12464830187622966,
      "learning_rate": 0.00015436726992505096,
      "loss": 1.1139,
      "step": 4683
    },
    {
      "epoch": 0.3376342535861025,
      "grad_norm": 0.16676449420936446,
      "learning_rate": 0.00015434767324205422,
      "loss": 1.0131,
      "step": 4684
    },
    {
      "epoch": 0.3377063360484394,
      "grad_norm": 0.13450293648860268,
      "learning_rate": 0.00015432807359662304,
      "loss": 1.0275,
      "step": 4685
    },
    {
      "epoch": 0.33777841851077633,
      "grad_norm": 0.12268096036623402,
      "learning_rate": 0.00015430847098982584,
      "loss": 0.9796,
      "step": 4686
    },
    {
      "epoch": 0.33785050097311325,
      "grad_norm": 0.14278574799889843,
      "learning_rate": 0.0001542888654227311,
      "loss": 1.1355,
      "step": 4687
    },
    {
      "epoch": 0.3379225834354502,
      "grad_norm": 0.21909481157631355,
      "learning_rate": 0.00015426925689640756,
      "loss": 1.0938,
      "step": 4688
    },
    {
      "epoch": 0.3379946658977871,
      "grad_norm": 0.1650173555610938,
      "learning_rate": 0.00015424964541192399,
      "loss": 1.0897,
      "step": 4689
    },
    {
      "epoch": 0.33806674836012396,
      "grad_norm": 0.18086059891210568,
      "learning_rate": 0.0001542300309703494,
      "loss": 1.106,
      "step": 4690
    },
    {
      "epoch": 0.3381388308224609,
      "grad_norm": 0.15882928984656078,
      "learning_rate": 0.00015421041357275298,
      "loss": 1.0573,
      "step": 4691
    },
    {
      "epoch": 0.3382109132847978,
      "grad_norm": 0.4440246320736768,
      "learning_rate": 0.00015419079322020402,
      "loss": 1.111,
      "step": 4692
    },
    {
      "epoch": 0.3382829957471347,
      "grad_norm": 0.1215481888030947,
      "learning_rate": 0.00015417116991377203,
      "loss": 1.0562,
      "step": 4693
    },
    {
      "epoch": 0.33835507820947164,
      "grad_norm": 0.11542493708973614,
      "learning_rate": 0.00015415154365452664,
      "loss": 1.0598,
      "step": 4694
    },
    {
      "epoch": 0.33842716067180856,
      "grad_norm": 0.13587904631141734,
      "learning_rate": 0.0001541319144435377,
      "loss": 1.1012,
      "step": 4695
    },
    {
      "epoch": 0.3384992431341455,
      "grad_norm": 1.9061752536995242,
      "learning_rate": 0.0001541122822818751,
      "loss": 1.0903,
      "step": 4696
    },
    {
      "epoch": 0.3385713255964824,
      "grad_norm": 0.1141053900599623,
      "learning_rate": 0.00015409264717060906,
      "loss": 1.0511,
      "step": 4697
    },
    {
      "epoch": 0.33864340805881926,
      "grad_norm": 0.17161680830993936,
      "learning_rate": 0.0001540730091108098,
      "loss": 1.0192,
      "step": 4698
    },
    {
      "epoch": 0.3387154905211562,
      "grad_norm": 2.5634711348534607,
      "learning_rate": 0.0001540533681035478,
      "loss": 1.0982,
      "step": 4699
    },
    {
      "epoch": 0.3387875729834931,
      "grad_norm": 0.158704151627992,
      "learning_rate": 0.00015403372414989368,
      "loss": 1.0533,
      "step": 4700
    },
    {
      "epoch": 0.33885965544583,
      "grad_norm": 0.10577369862700896,
      "learning_rate": 0.00015401407725091815,
      "loss": 1.0465,
      "step": 4701
    },
    {
      "epoch": 0.33893173790816694,
      "grad_norm": 0.1279862156109364,
      "learning_rate": 0.0001539944274076922,
      "loss": 1.006,
      "step": 4702
    },
    {
      "epoch": 0.33900382037050386,
      "grad_norm": 0.13498014963900137,
      "learning_rate": 0.00015397477462128693,
      "loss": 1.117,
      "step": 4703
    },
    {
      "epoch": 0.3390759028328408,
      "grad_norm": 0.21749241862848112,
      "learning_rate": 0.00015395511889277355,
      "loss": 1.1414,
      "step": 4704
    },
    {
      "epoch": 0.3391479852951777,
      "grad_norm": 0.12481243515262189,
      "learning_rate": 0.00015393546022322353,
      "loss": 1.1021,
      "step": 4705
    },
    {
      "epoch": 0.3392200677575146,
      "grad_norm": 0.3458776814663437,
      "learning_rate": 0.00015391579861370836,
      "loss": 1.0622,
      "step": 4706
    },
    {
      "epoch": 0.3392921502198515,
      "grad_norm": 0.12031948706034692,
      "learning_rate": 0.00015389613406529984,
      "loss": 0.9917,
      "step": 4707
    },
    {
      "epoch": 0.3393642326821884,
      "grad_norm": 0.16525198957412154,
      "learning_rate": 0.00015387646657906987,
      "loss": 1.045,
      "step": 4708
    },
    {
      "epoch": 0.3394363151445253,
      "grad_norm": 1.541337714912897,
      "learning_rate": 0.00015385679615609042,
      "loss": 1.0979,
      "step": 4709
    },
    {
      "epoch": 0.33950839760686224,
      "grad_norm": 0.1585464511817358,
      "learning_rate": 0.0001538371227974338,
      "loss": 1.1385,
      "step": 4710
    },
    {
      "epoch": 0.33958048006919916,
      "grad_norm": 0.15753821928889403,
      "learning_rate": 0.0001538174465041724,
      "loss": 1.0262,
      "step": 4711
    },
    {
      "epoch": 0.3396525625315361,
      "grad_norm": 0.16991709301702632,
      "learning_rate": 0.0001537977672773786,
      "loss": 1.0484,
      "step": 4712
    },
    {
      "epoch": 0.339724644993873,
      "grad_norm": 0.48370914380593827,
      "learning_rate": 0.00015377808511812525,
      "loss": 1.0581,
      "step": 4713
    },
    {
      "epoch": 0.3397967274562099,
      "grad_norm": 0.14976383375310942,
      "learning_rate": 0.0001537584000274851,
      "loss": 1.0439,
      "step": 4714
    },
    {
      "epoch": 0.33986880991854684,
      "grad_norm": 0.11615813335013604,
      "learning_rate": 0.00015373871200653131,
      "loss": 1.0276,
      "step": 4715
    },
    {
      "epoch": 0.3399408923808837,
      "grad_norm": 0.168012771733113,
      "learning_rate": 0.00015371902105633687,
      "loss": 1.0536,
      "step": 4716
    },
    {
      "epoch": 0.34001297484322063,
      "grad_norm": 0.19643732213381263,
      "learning_rate": 0.00015369932717797523,
      "loss": 1.0475,
      "step": 4717
    },
    {
      "epoch": 0.34008505730555755,
      "grad_norm": 1.1233441561841266,
      "learning_rate": 0.00015367963037251983,
      "loss": 1.1049,
      "step": 4718
    },
    {
      "epoch": 0.34015713976789447,
      "grad_norm": 0.19933713138791234,
      "learning_rate": 0.00015365993064104437,
      "loss": 1.1013,
      "step": 4719
    },
    {
      "epoch": 0.3402292222302314,
      "grad_norm": 0.12307125605649445,
      "learning_rate": 0.0001536402279846226,
      "loss": 1.1088,
      "step": 4720
    },
    {
      "epoch": 0.3403013046925683,
      "grad_norm": 0.19319009421774025,
      "learning_rate": 0.00015362052240432857,
      "loss": 1.0406,
      "step": 4721
    },
    {
      "epoch": 0.3403733871549052,
      "grad_norm": 0.10915593484207109,
      "learning_rate": 0.00015360081390123639,
      "loss": 1.012,
      "step": 4722
    },
    {
      "epoch": 0.34044546961724215,
      "grad_norm": 0.18241888447030172,
      "learning_rate": 0.00015358110247642025,
      "loss": 1.0475,
      "step": 4723
    },
    {
      "epoch": 0.340517552079579,
      "grad_norm": 0.8884127566095147,
      "learning_rate": 0.0001535613881309547,
      "loss": 1.0716,
      "step": 4724
    },
    {
      "epoch": 0.34058963454191593,
      "grad_norm": 0.14856633826271515,
      "learning_rate": 0.00015354167086591432,
      "loss": 1.0619,
      "step": 4725
    },
    {
      "epoch": 0.34066171700425285,
      "grad_norm": 0.10869970285623365,
      "learning_rate": 0.0001535219506823739,
      "loss": 1.0611,
      "step": 4726
    },
    {
      "epoch": 0.34073379946658977,
      "grad_norm": 0.18377001173889915,
      "learning_rate": 0.00015350222758140833,
      "loss": 1.1389,
      "step": 4727
    },
    {
      "epoch": 0.3408058819289267,
      "grad_norm": 0.13057484159355326,
      "learning_rate": 0.00015348250156409273,
      "loss": 1.1338,
      "step": 4728
    },
    {
      "epoch": 0.3408779643912636,
      "grad_norm": 0.1995844569401374,
      "learning_rate": 0.0001534627726315023,
      "loss": 1.0485,
      "step": 4729
    },
    {
      "epoch": 0.34095004685360053,
      "grad_norm": 0.11200394672871881,
      "learning_rate": 0.0001534430407847125,
      "loss": 1.0248,
      "step": 4730
    },
    {
      "epoch": 0.34102212931593745,
      "grad_norm": 0.4968431072340866,
      "learning_rate": 0.00015342330602479884,
      "loss": 1.0288,
      "step": 4731
    },
    {
      "epoch": 0.34109421177827437,
      "grad_norm": 0.12718497878911075,
      "learning_rate": 0.00015340356835283708,
      "loss": 1.0892,
      "step": 4732
    },
    {
      "epoch": 0.34116629424061123,
      "grad_norm": 0.12225285889926658,
      "learning_rate": 0.00015338382776990307,
      "loss": 1.124,
      "step": 4733
    },
    {
      "epoch": 0.34123837670294815,
      "grad_norm": 0.10727836294037639,
      "learning_rate": 0.00015336408427707286,
      "loss": 0.9748,
      "step": 4734
    },
    {
      "epoch": 0.3413104591652851,
      "grad_norm": 0.14164417529820994,
      "learning_rate": 0.00015334433787542266,
      "loss": 1.1402,
      "step": 4735
    },
    {
      "epoch": 0.341382541627622,
      "grad_norm": 0.14261927496784157,
      "learning_rate": 0.00015332458856602877,
      "loss": 1.1501,
      "step": 4736
    },
    {
      "epoch": 0.3414546240899589,
      "grad_norm": 0.10807151777605985,
      "learning_rate": 0.00015330483634996782,
      "loss": 1.0659,
      "step": 4737
    },
    {
      "epoch": 0.34152670655229583,
      "grad_norm": 0.1220319466979688,
      "learning_rate": 0.00015328508122831636,
      "loss": 1.0466,
      "step": 4738
    },
    {
      "epoch": 0.34159878901463275,
      "grad_norm": 0.11697660278081377,
      "learning_rate": 0.0001532653232021513,
      "loss": 1.1075,
      "step": 4739
    },
    {
      "epoch": 0.3416708714769697,
      "grad_norm": 0.11455480385737356,
      "learning_rate": 0.00015324556227254958,
      "loss": 1.0749,
      "step": 4740
    },
    {
      "epoch": 0.3417429539393066,
      "grad_norm": 0.14696349190237323,
      "learning_rate": 0.00015322579844058838,
      "loss": 1.076,
      "step": 4741
    },
    {
      "epoch": 0.34181503640164346,
      "grad_norm": 0.13982702064081481,
      "learning_rate": 0.00015320603170734497,
      "loss": 1.0629,
      "step": 4742
    },
    {
      "epoch": 0.3418871188639804,
      "grad_norm": 0.13495912007937647,
      "learning_rate": 0.00015318626207389685,
      "loss": 1.0392,
      "step": 4743
    },
    {
      "epoch": 0.3419592013263173,
      "grad_norm": 0.12611739765188112,
      "learning_rate": 0.00015316648954132164,
      "loss": 1.0433,
      "step": 4744
    },
    {
      "epoch": 0.3420312837886542,
      "grad_norm": 0.34392730882101263,
      "learning_rate": 0.00015314671411069712,
      "loss": 1.0478,
      "step": 4745
    },
    {
      "epoch": 0.34210336625099114,
      "grad_norm": 0.11790003044183303,
      "learning_rate": 0.0001531269357831012,
      "loss": 1.0743,
      "step": 4746
    },
    {
      "epoch": 0.34217544871332806,
      "grad_norm": 0.150507998168778,
      "learning_rate": 0.00015310715455961197,
      "loss": 1.0027,
      "step": 4747
    },
    {
      "epoch": 0.342247531175665,
      "grad_norm": 0.29990391734342897,
      "learning_rate": 0.00015308737044130775,
      "loss": 0.9873,
      "step": 4748
    },
    {
      "epoch": 0.3423196136380019,
      "grad_norm": 0.33753306458362636,
      "learning_rate": 0.00015306758342926687,
      "loss": 1.0775,
      "step": 4749
    },
    {
      "epoch": 0.34239169610033876,
      "grad_norm": 0.13342649230516127,
      "learning_rate": 0.00015304779352456795,
      "loss": 1.1041,
      "step": 4750
    },
    {
      "epoch": 0.3424637785626757,
      "grad_norm": 0.12367680868806223,
      "learning_rate": 0.0001530280007282897,
      "loss": 1.1165,
      "step": 4751
    },
    {
      "epoch": 0.3425358610250126,
      "grad_norm": 0.4609336048457374,
      "learning_rate": 0.000153008205041511,
      "loss": 1.1601,
      "step": 4752
    },
    {
      "epoch": 0.3426079434873495,
      "grad_norm": 0.1028096413773031,
      "learning_rate": 0.0001529884064653109,
      "loss": 1.0557,
      "step": 4753
    },
    {
      "epoch": 0.34268002594968644,
      "grad_norm": 0.1561317802997789,
      "learning_rate": 0.00015296860500076863,
      "loss": 1.0564,
      "step": 4754
    },
    {
      "epoch": 0.34275210841202336,
      "grad_norm": 0.1240878518052694,
      "learning_rate": 0.00015294880064896352,
      "loss": 1.0042,
      "step": 4755
    },
    {
      "epoch": 0.3428241908743603,
      "grad_norm": 0.4988315042039233,
      "learning_rate": 0.00015292899341097505,
      "loss": 1.098,
      "step": 4756
    },
    {
      "epoch": 0.3428962733366972,
      "grad_norm": 0.1994615548502869,
      "learning_rate": 0.00015290918328788298,
      "loss": 0.9781,
      "step": 4757
    },
    {
      "epoch": 0.3429683557990341,
      "grad_norm": 0.12054398024861242,
      "learning_rate": 0.000152889370280767,
      "loss": 1.0527,
      "step": 4758
    },
    {
      "epoch": 0.343040438261371,
      "grad_norm": 0.14678167905551728,
      "learning_rate": 0.00015286955439070723,
      "loss": 1.0693,
      "step": 4759
    },
    {
      "epoch": 0.3431125207237079,
      "grad_norm": 0.24803929929292823,
      "learning_rate": 0.00015284973561878373,
      "loss": 1.1304,
      "step": 4760
    },
    {
      "epoch": 0.3431846031860448,
      "grad_norm": 0.2801553353119536,
      "learning_rate": 0.00015282991396607686,
      "loss": 1.0782,
      "step": 4761
    },
    {
      "epoch": 0.34325668564838174,
      "grad_norm": 0.11991898630996159,
      "learning_rate": 0.000152810089433667,
      "loss": 1.1946,
      "step": 4762
    },
    {
      "epoch": 0.34332876811071866,
      "grad_norm": 2.540299354839368,
      "learning_rate": 0.00015279026202263488,
      "loss": 1.0217,
      "step": 4763
    },
    {
      "epoch": 0.3434008505730556,
      "grad_norm": 0.11164589454708682,
      "learning_rate": 0.00015277043173406117,
      "loss": 1.0148,
      "step": 4764
    },
    {
      "epoch": 0.3434729330353925,
      "grad_norm": 0.26108056544964114,
      "learning_rate": 0.00015275059856902684,
      "loss": 1.0874,
      "step": 4765
    },
    {
      "epoch": 0.3435450154977294,
      "grad_norm": 0.12271941744599515,
      "learning_rate": 0.00015273076252861296,
      "loss": 1.0585,
      "step": 4766
    },
    {
      "epoch": 0.34361709796006634,
      "grad_norm": 0.1464138580740618,
      "learning_rate": 0.00015271092361390077,
      "loss": 1.0705,
      "step": 4767
    },
    {
      "epoch": 0.3436891804224032,
      "grad_norm": 0.23883767343642012,
      "learning_rate": 0.00015269108182597172,
      "loss": 1.0192,
      "step": 4768
    },
    {
      "epoch": 0.3437612628847401,
      "grad_norm": 0.2510819008729765,
      "learning_rate": 0.00015267123716590728,
      "loss": 1.0821,
      "step": 4769
    },
    {
      "epoch": 0.34383334534707705,
      "grad_norm": 0.11864947629857084,
      "learning_rate": 0.0001526513896347892,
      "loss": 1.0608,
      "step": 4770
    },
    {
      "epoch": 0.34390542780941397,
      "grad_norm": 0.19820749755648898,
      "learning_rate": 0.00015263153923369938,
      "loss": 1.0263,
      "step": 4771
    },
    {
      "epoch": 0.3439775102717509,
      "grad_norm": 0.11861688669589615,
      "learning_rate": 0.0001526116859637198,
      "loss": 1.0368,
      "step": 4772
    },
    {
      "epoch": 0.3440495927340878,
      "grad_norm": 0.11657406141309927,
      "learning_rate": 0.0001525918298259327,
      "loss": 1.0246,
      "step": 4773
    },
    {
      "epoch": 0.3441216751964247,
      "grad_norm": 0.12466298089204177,
      "learning_rate": 0.00015257197082142032,
      "loss": 1.078,
      "step": 4774
    },
    {
      "epoch": 0.34419375765876165,
      "grad_norm": 0.1252944076667136,
      "learning_rate": 0.00015255210895126525,
      "loss": 1.093,
      "step": 4775
    },
    {
      "epoch": 0.3442658401210985,
      "grad_norm": 0.1325935206818598,
      "learning_rate": 0.00015253224421655004,
      "loss": 1.0343,
      "step": 4776
    },
    {
      "epoch": 0.34433792258343543,
      "grad_norm": 0.16259598796947217,
      "learning_rate": 0.0001525123766183576,
      "loss": 1.0396,
      "step": 4777
    },
    {
      "epoch": 0.34441000504577235,
      "grad_norm": 0.18528229413726785,
      "learning_rate": 0.00015249250615777084,
      "loss": 0.9806,
      "step": 4778
    },
    {
      "epoch": 0.34448208750810927,
      "grad_norm": 0.18601563316061628,
      "learning_rate": 0.00015247263283587286,
      "loss": 1.0606,
      "step": 4779
    },
    {
      "epoch": 0.3445541699704462,
      "grad_norm": 0.3795513214254639,
      "learning_rate": 0.00015245275665374698,
      "loss": 1.0241,
      "step": 4780
    },
    {
      "epoch": 0.3446262524327831,
      "grad_norm": 0.22506606286194245,
      "learning_rate": 0.0001524328776124766,
      "loss": 1.1075,
      "step": 4781
    },
    {
      "epoch": 0.34469833489512003,
      "grad_norm": 0.12453196742405759,
      "learning_rate": 0.0001524129957131453,
      "loss": 1.0495,
      "step": 4782
    },
    {
      "epoch": 0.34477041735745695,
      "grad_norm": 0.11434243496436335,
      "learning_rate": 0.00015239311095683683,
      "loss": 1.0251,
      "step": 4783
    },
    {
      "epoch": 0.34484249981979387,
      "grad_norm": 0.12182383942265711,
      "learning_rate": 0.0001523732233446351,
      "loss": 1.049,
      "step": 4784
    },
    {
      "epoch": 0.34491458228213073,
      "grad_norm": 0.2347303472110709,
      "learning_rate": 0.00015235333287762418,
      "loss": 1.0414,
      "step": 4785
    },
    {
      "epoch": 0.34498666474446765,
      "grad_norm": 0.20493281419630205,
      "learning_rate": 0.00015233343955688822,
      "loss": 1.0349,
      "step": 4786
    },
    {
      "epoch": 0.3450587472068046,
      "grad_norm": 0.12853758108584298,
      "learning_rate": 0.00015231354338351164,
      "loss": 1.0736,
      "step": 4787
    },
    {
      "epoch": 0.3451308296691415,
      "grad_norm": 0.11483621084110851,
      "learning_rate": 0.00015229364435857893,
      "loss": 0.9969,
      "step": 4788
    },
    {
      "epoch": 0.3452029121314784,
      "grad_norm": 0.16267679292555998,
      "learning_rate": 0.00015227374248317474,
      "loss": 1.055,
      "step": 4789
    },
    {
      "epoch": 0.34527499459381533,
      "grad_norm": 0.2184318895586677,
      "learning_rate": 0.00015225383775838397,
      "loss": 1.045,
      "step": 4790
    },
    {
      "epoch": 0.34534707705615225,
      "grad_norm": 0.18328714033868695,
      "learning_rate": 0.00015223393018529152,
      "loss": 1.0785,
      "step": 4791
    },
    {
      "epoch": 0.34541915951848917,
      "grad_norm": 0.16345071506028364,
      "learning_rate": 0.00015221401976498264,
      "loss": 1.1189,
      "step": 4792
    },
    {
      "epoch": 0.3454912419808261,
      "grad_norm": 0.12172434732638651,
      "learning_rate": 0.00015219410649854252,
      "loss": 1.0554,
      "step": 4793
    },
    {
      "epoch": 0.34556332444316296,
      "grad_norm": 0.12659513501321276,
      "learning_rate": 0.00015217419038705666,
      "loss": 1.1339,
      "step": 4794
    },
    {
      "epoch": 0.3456354069054999,
      "grad_norm": 0.12662223747074652,
      "learning_rate": 0.00015215427143161066,
      "loss": 1.0889,
      "step": 4795
    },
    {
      "epoch": 0.3457074893678368,
      "grad_norm": 0.3676074478586541,
      "learning_rate": 0.0001521343496332903,
      "loss": 1.0998,
      "step": 4796
    },
    {
      "epoch": 0.3457795718301737,
      "grad_norm": 0.12931466941775796,
      "learning_rate": 0.00015211442499318145,
      "loss": 1.0774,
      "step": 4797
    },
    {
      "epoch": 0.34585165429251064,
      "grad_norm": 0.17327391407402923,
      "learning_rate": 0.00015209449751237025,
      "loss": 1.0436,
      "step": 4798
    },
    {
      "epoch": 0.34592373675484756,
      "grad_norm": 0.1135260524288423,
      "learning_rate": 0.00015207456719194284,
      "loss": 1.1357,
      "step": 4799
    },
    {
      "epoch": 0.3459958192171845,
      "grad_norm": 0.11733215816957128,
      "learning_rate": 0.00015205463403298566,
      "loss": 1.0287,
      "step": 4800
    },
    {
      "epoch": 0.3460679016795214,
      "grad_norm": 0.14159664879833767,
      "learning_rate": 0.00015203469803658521,
      "loss": 1.0716,
      "step": 4801
    },
    {
      "epoch": 0.34613998414185826,
      "grad_norm": 0.2125838339193271,
      "learning_rate": 0.00015201475920382823,
      "loss": 1.0301,
      "step": 4802
    },
    {
      "epoch": 0.3462120666041952,
      "grad_norm": 0.2420566684932982,
      "learning_rate": 0.00015199481753580153,
      "loss": 1.0477,
      "step": 4803
    },
    {
      "epoch": 0.3462841490665321,
      "grad_norm": 0.2431550490824497,
      "learning_rate": 0.00015197487303359212,
      "loss": 1.1149,
      "step": 4804
    },
    {
      "epoch": 0.346356231528869,
      "grad_norm": 0.12826052191736284,
      "learning_rate": 0.00015195492569828716,
      "loss": 1.1245,
      "step": 4805
    },
    {
      "epoch": 0.34642831399120594,
      "grad_norm": 0.20220358190150822,
      "learning_rate": 0.0001519349755309739,
      "loss": 1.0002,
      "step": 4806
    },
    {
      "epoch": 0.34650039645354286,
      "grad_norm": 0.11811466464466765,
      "learning_rate": 0.0001519150225327399,
      "loss": 0.9792,
      "step": 4807
    },
    {
      "epoch": 0.3465724789158798,
      "grad_norm": 0.11537647050482872,
      "learning_rate": 0.00015189506670467272,
      "loss": 1.1242,
      "step": 4808
    },
    {
      "epoch": 0.3466445613782167,
      "grad_norm": 0.278361157205508,
      "learning_rate": 0.00015187510804786012,
      "loss": 1.0487,
      "step": 4809
    },
    {
      "epoch": 0.3467166438405536,
      "grad_norm": 0.171069737660319,
      "learning_rate": 0.00015185514656339006,
      "loss": 1.0677,
      "step": 4810
    },
    {
      "epoch": 0.3467887263028905,
      "grad_norm": 0.12060446964605534,
      "learning_rate": 0.0001518351822523506,
      "loss": 1.0373,
      "step": 4811
    },
    {
      "epoch": 0.3468608087652274,
      "grad_norm": 0.10612292431902408,
      "learning_rate": 0.00015181521511582996,
      "loss": 1.0036,
      "step": 4812
    },
    {
      "epoch": 0.3469328912275643,
      "grad_norm": 0.26375236525576135,
      "learning_rate": 0.00015179524515491656,
      "loss": 1.1003,
      "step": 4813
    },
    {
      "epoch": 0.34700497368990124,
      "grad_norm": 0.11306946084672269,
      "learning_rate": 0.0001517752723706989,
      "loss": 1.1293,
      "step": 4814
    },
    {
      "epoch": 0.34707705615223816,
      "grad_norm": 0.13194200152354732,
      "learning_rate": 0.0001517552967642657,
      "loss": 0.9819,
      "step": 4815
    },
    {
      "epoch": 0.3471491386145751,
      "grad_norm": 0.6205231541439851,
      "learning_rate": 0.00015173531833670586,
      "loss": 1.0843,
      "step": 4816
    },
    {
      "epoch": 0.347221221076912,
      "grad_norm": 0.1336534991664611,
      "learning_rate": 0.00015171533708910827,
      "loss": 1.0929,
      "step": 4817
    },
    {
      "epoch": 0.3472933035392489,
      "grad_norm": 0.12390691510769618,
      "learning_rate": 0.00015169535302256222,
      "loss": 1.0661,
      "step": 4818
    },
    {
      "epoch": 0.34736538600158584,
      "grad_norm": 0.10216894211942357,
      "learning_rate": 0.00015167536613815692,
      "loss": 1.0656,
      "step": 4819
    },
    {
      "epoch": 0.3474374684639227,
      "grad_norm": 0.14830060439693765,
      "learning_rate": 0.00015165537643698184,
      "loss": 1.0776,
      "step": 4820
    },
    {
      "epoch": 0.3475095509262596,
      "grad_norm": 0.11116445062422654,
      "learning_rate": 0.00015163538392012663,
      "loss": 1.0045,
      "step": 4821
    },
    {
      "epoch": 0.34758163338859654,
      "grad_norm": 0.3473977712130543,
      "learning_rate": 0.00015161538858868108,
      "loss": 0.9779,
      "step": 4822
    },
    {
      "epoch": 0.34765371585093346,
      "grad_norm": 0.12728163480587595,
      "learning_rate": 0.00015159539044373504,
      "loss": 1.1225,
      "step": 4823
    },
    {
      "epoch": 0.3477257983132704,
      "grad_norm": 0.10279693946462809,
      "learning_rate": 0.00015157538948637868,
      "loss": 1.0689,
      "step": 4824
    },
    {
      "epoch": 0.3477978807756073,
      "grad_norm": 1.6263990792158471,
      "learning_rate": 0.00015155538571770218,
      "loss": 0.9441,
      "step": 4825
    },
    {
      "epoch": 0.3478699632379442,
      "grad_norm": 0.12001067994318018,
      "learning_rate": 0.00015153537913879594,
      "loss": 1.0796,
      "step": 4826
    },
    {
      "epoch": 0.34794204570028114,
      "grad_norm": 0.2683503505486029,
      "learning_rate": 0.00015151536975075046,
      "loss": 1.123,
      "step": 4827
    },
    {
      "epoch": 0.348014128162618,
      "grad_norm": 0.1781729181596812,
      "learning_rate": 0.00015149535755465648,
      "loss": 1.0852,
      "step": 4828
    },
    {
      "epoch": 0.34808621062495493,
      "grad_norm": 0.46560612318004907,
      "learning_rate": 0.0001514753425516048,
      "loss": 1.104,
      "step": 4829
    },
    {
      "epoch": 0.34815829308729185,
      "grad_norm": 0.14825298597389056,
      "learning_rate": 0.00015145532474268646,
      "loss": 1.0452,
      "step": 4830
    },
    {
      "epoch": 0.34823037554962877,
      "grad_norm": 0.10590632590028717,
      "learning_rate": 0.00015143530412899255,
      "loss": 1.0654,
      "step": 4831
    },
    {
      "epoch": 0.3483024580119657,
      "grad_norm": 0.12667477825387335,
      "learning_rate": 0.00015141528071161448,
      "loss": 1.0925,
      "step": 4832
    },
    {
      "epoch": 0.3483745404743026,
      "grad_norm": 0.11415429931321587,
      "learning_rate": 0.00015139525449164358,
      "loss": 1.0644,
      "step": 4833
    },
    {
      "epoch": 0.3484466229366395,
      "grad_norm": 0.1155871754555398,
      "learning_rate": 0.00015137522547017157,
      "loss": 1.1439,
      "step": 4834
    },
    {
      "epoch": 0.34851870539897645,
      "grad_norm": 0.11460696965624749,
      "learning_rate": 0.00015135519364829007,
      "loss": 1.0667,
      "step": 4835
    },
    {
      "epoch": 0.34859078786131337,
      "grad_norm": 0.15748865650442626,
      "learning_rate": 0.00015133515902709116,
      "loss": 1.0524,
      "step": 4836
    },
    {
      "epoch": 0.34866287032365023,
      "grad_norm": 0.16957347867744904,
      "learning_rate": 0.0001513151216076668,
      "loss": 1.1341,
      "step": 4837
    },
    {
      "epoch": 0.34873495278598715,
      "grad_norm": 0.12376869339670646,
      "learning_rate": 0.00015129508139110923,
      "loss": 1.0433,
      "step": 4838
    },
    {
      "epoch": 0.34880703524832407,
      "grad_norm": 0.1291418227904641,
      "learning_rate": 0.0001512750383785108,
      "loss": 1.0496,
      "step": 4839
    },
    {
      "epoch": 0.348879117710661,
      "grad_norm": 0.2705792500053715,
      "learning_rate": 0.0001512549925709641,
      "loss": 1.0248,
      "step": 4840
    },
    {
      "epoch": 0.3489512001729979,
      "grad_norm": 0.16810288991132663,
      "learning_rate": 0.00015123494396956176,
      "loss": 1.0643,
      "step": 4841
    },
    {
      "epoch": 0.34902328263533483,
      "grad_norm": 0.16272356210185845,
      "learning_rate": 0.00015121489257539662,
      "loss": 1.0811,
      "step": 4842
    },
    {
      "epoch": 0.34909536509767175,
      "grad_norm": 0.27151626884035696,
      "learning_rate": 0.0001511948383895616,
      "loss": 1.0671,
      "step": 4843
    },
    {
      "epoch": 0.34916744756000867,
      "grad_norm": 0.1149513306291732,
      "learning_rate": 0.00015117478141314994,
      "loss": 1.0238,
      "step": 4844
    },
    {
      "epoch": 0.3492395300223456,
      "grad_norm": 0.14003265008753132,
      "learning_rate": 0.00015115472164725486,
      "loss": 1.0506,
      "step": 4845
    },
    {
      "epoch": 0.34931161248468245,
      "grad_norm": 0.10896163917215869,
      "learning_rate": 0.00015113465909296977,
      "loss": 1.0771,
      "step": 4846
    },
    {
      "epoch": 0.3493836949470194,
      "grad_norm": 0.11618546471761386,
      "learning_rate": 0.00015111459375138835,
      "loss": 1.1128,
      "step": 4847
    },
    {
      "epoch": 0.3494557774093563,
      "grad_norm": 0.1032178243112307,
      "learning_rate": 0.00015109452562360424,
      "loss": 1.0957,
      "step": 4848
    },
    {
      "epoch": 0.3495278598716932,
      "grad_norm": 0.09876483570548557,
      "learning_rate": 0.0001510744547107114,
      "loss": 1.0794,
      "step": 4849
    },
    {
      "epoch": 0.34959994233403013,
      "grad_norm": 0.18501351485807524,
      "learning_rate": 0.00015105438101380383,
      "loss": 1.1122,
      "step": 4850
    },
    {
      "epoch": 0.34967202479636705,
      "grad_norm": 0.11737530081939831,
      "learning_rate": 0.0001510343045339758,
      "loss": 1.0243,
      "step": 4851
    },
    {
      "epoch": 0.349744107258704,
      "grad_norm": 0.18972231658141572,
      "learning_rate": 0.00015101422527232156,
      "loss": 1.0313,
      "step": 4852
    },
    {
      "epoch": 0.3498161897210409,
      "grad_norm": 0.10518849972032972,
      "learning_rate": 0.00015099414322993567,
      "loss": 1.1207,
      "step": 4853
    },
    {
      "epoch": 0.34988827218337776,
      "grad_norm": 0.107223976787674,
      "learning_rate": 0.00015097405840791276,
      "loss": 1.0545,
      "step": 4854
    },
    {
      "epoch": 0.3499603546457147,
      "grad_norm": 0.11145434777216918,
      "learning_rate": 0.00015095397080734768,
      "loss": 1.0752,
      "step": 4855
    },
    {
      "epoch": 0.3500324371080516,
      "grad_norm": 0.11867971187941409,
      "learning_rate": 0.0001509338804293353,
      "loss": 1.0847,
      "step": 4856
    },
    {
      "epoch": 0.3501045195703885,
      "grad_norm": 0.2613605749493613,
      "learning_rate": 0.0001509137872749708,
      "loss": 1.0474,
      "step": 4857
    },
    {
      "epoch": 0.35017660203272544,
      "grad_norm": 0.1125874052931899,
      "learning_rate": 0.00015089369134534938,
      "loss": 1.0058,
      "step": 4858
    },
    {
      "epoch": 0.35024868449506236,
      "grad_norm": 0.11571011301377869,
      "learning_rate": 0.00015087359264156648,
      "loss": 1.0875,
      "step": 4859
    },
    {
      "epoch": 0.3503207669573993,
      "grad_norm": 0.3121290565990109,
      "learning_rate": 0.00015085349116471765,
      "loss": 1.0933,
      "step": 4860
    },
    {
      "epoch": 0.3503928494197362,
      "grad_norm": 0.33904121377543034,
      "learning_rate": 0.0001508333869158986,
      "loss": 1.0676,
      "step": 4861
    },
    {
      "epoch": 0.3504649318820731,
      "grad_norm": 0.10175098637657308,
      "learning_rate": 0.00015081327989620522,
      "loss": 1.0058,
      "step": 4862
    },
    {
      "epoch": 0.35053701434441,
      "grad_norm": 0.10768690977949168,
      "learning_rate": 0.0001507931701067335,
      "loss": 1.066,
      "step": 4863
    },
    {
      "epoch": 0.3506090968067469,
      "grad_norm": 0.12337409873693028,
      "learning_rate": 0.00015077305754857955,
      "loss": 1.0021,
      "step": 4864
    },
    {
      "epoch": 0.3506811792690838,
      "grad_norm": 0.10568641632856003,
      "learning_rate": 0.0001507529422228398,
      "loss": 1.0487,
      "step": 4865
    },
    {
      "epoch": 0.35075326173142074,
      "grad_norm": 0.1297047161959608,
      "learning_rate": 0.00015073282413061062,
      "loss": 1.1042,
      "step": 4866
    },
    {
      "epoch": 0.35082534419375766,
      "grad_norm": 0.10715073844015109,
      "learning_rate": 0.00015071270327298865,
      "loss": 1.0653,
      "step": 4867
    },
    {
      "epoch": 0.3508974266560946,
      "grad_norm": 0.1258848125963353,
      "learning_rate": 0.00015069257965107065,
      "loss": 0.9853,
      "step": 4868
    },
    {
      "epoch": 0.3509695091184315,
      "grad_norm": 0.10662285471876994,
      "learning_rate": 0.00015067245326595358,
      "loss": 1.0185,
      "step": 4869
    },
    {
      "epoch": 0.3510415915807684,
      "grad_norm": 0.14639490491777923,
      "learning_rate": 0.00015065232411873444,
      "loss": 1.0649,
      "step": 4870
    },
    {
      "epoch": 0.35111367404310534,
      "grad_norm": 0.13995283027613836,
      "learning_rate": 0.00015063219221051055,
      "loss": 1.0017,
      "step": 4871
    },
    {
      "epoch": 0.3511857565054422,
      "grad_norm": 0.12962001121733163,
      "learning_rate": 0.00015061205754237915,
      "loss": 1.0416,
      "step": 4872
    },
    {
      "epoch": 0.3512578389677791,
      "grad_norm": 0.14221796595284053,
      "learning_rate": 0.00015059192011543786,
      "loss": 1.0209,
      "step": 4873
    },
    {
      "epoch": 0.35132992143011604,
      "grad_norm": 0.11036397519243887,
      "learning_rate": 0.0001505717799307843,
      "loss": 1.0429,
      "step": 4874
    },
    {
      "epoch": 0.35140200389245296,
      "grad_norm": 0.10673706821940927,
      "learning_rate": 0.0001505516369895163,
      "loss": 1.0373,
      "step": 4875
    },
    {
      "epoch": 0.3514740863547899,
      "grad_norm": 1.7403525105079944,
      "learning_rate": 0.00015053149129273186,
      "loss": 1.0132,
      "step": 4876
    },
    {
      "epoch": 0.3515461688171268,
      "grad_norm": 0.1591401160611141,
      "learning_rate": 0.00015051134284152907,
      "loss": 1.1003,
      "step": 4877
    },
    {
      "epoch": 0.3516182512794637,
      "grad_norm": 0.14749418439296735,
      "learning_rate": 0.0001504911916370062,
      "loss": 1.0554,
      "step": 4878
    },
    {
      "epoch": 0.35169033374180064,
      "grad_norm": 0.1144429683119466,
      "learning_rate": 0.0001504710376802617,
      "loss": 1.0761,
      "step": 4879
    },
    {
      "epoch": 0.3517624162041375,
      "grad_norm": 0.36126705947084337,
      "learning_rate": 0.00015045088097239407,
      "loss": 1.1161,
      "step": 4880
    },
    {
      "epoch": 0.3518344986664744,
      "grad_norm": 0.12190034054691717,
      "learning_rate": 0.00015043072151450212,
      "loss": 1.1399,
      "step": 4881
    },
    {
      "epoch": 0.35190658112881135,
      "grad_norm": 0.13436139611968653,
      "learning_rate": 0.00015041055930768468,
      "loss": 1.1709,
      "step": 4882
    },
    {
      "epoch": 0.35197866359114827,
      "grad_norm": 0.10922301960887992,
      "learning_rate": 0.00015039039435304078,
      "loss": 1.0747,
      "step": 4883
    },
    {
      "epoch": 0.3520507460534852,
      "grad_norm": 0.13227043727489268,
      "learning_rate": 0.00015037022665166956,
      "loss": 1.1926,
      "step": 4884
    },
    {
      "epoch": 0.3521228285158221,
      "grad_norm": 0.12881125636433927,
      "learning_rate": 0.0001503500562046704,
      "loss": 1.0225,
      "step": 4885
    },
    {
      "epoch": 0.352194910978159,
      "grad_norm": 0.13648814284045846,
      "learning_rate": 0.00015032988301314268,
      "loss": 1.0827,
      "step": 4886
    },
    {
      "epoch": 0.35226699344049595,
      "grad_norm": 0.22987722033243999,
      "learning_rate": 0.00015030970707818612,
      "loss": 1.0548,
      "step": 4887
    },
    {
      "epoch": 0.35233907590283287,
      "grad_norm": 0.1097248554748653,
      "learning_rate": 0.0001502895284009004,
      "loss": 1.006,
      "step": 4888
    },
    {
      "epoch": 0.35241115836516973,
      "grad_norm": 0.12075190818201141,
      "learning_rate": 0.0001502693469823855,
      "loss": 1.056,
      "step": 4889
    },
    {
      "epoch": 0.35248324082750665,
      "grad_norm": 0.13033448045451151,
      "learning_rate": 0.00015024916282374147,
      "loss": 0.9688,
      "step": 4890
    },
    {
      "epoch": 0.35255532328984357,
      "grad_norm": 0.11292906759889715,
      "learning_rate": 0.00015022897592606853,
      "loss": 1.0441,
      "step": 4891
    },
    {
      "epoch": 0.3526274057521805,
      "grad_norm": 0.11356994283952714,
      "learning_rate": 0.000150208786290467,
      "loss": 1.0281,
      "step": 4892
    },
    {
      "epoch": 0.3526994882145174,
      "grad_norm": 0.11171041694047348,
      "learning_rate": 0.0001501885939180375,
      "loss": 1.0486,
      "step": 4893
    },
    {
      "epoch": 0.35277157067685433,
      "grad_norm": 2.3302757420394618,
      "learning_rate": 0.00015016839880988055,
      "loss": 1.1541,
      "step": 4894
    },
    {
      "epoch": 0.35284365313919125,
      "grad_norm": 1.0415488302773084,
      "learning_rate": 0.0001501482009670971,
      "loss": 1.1767,
      "step": 4895
    },
    {
      "epoch": 0.35291573560152817,
      "grad_norm": 0.5718963074379404,
      "learning_rate": 0.00015012800039078807,
      "loss": 1.0909,
      "step": 4896
    },
    {
      "epoch": 0.3529878180638651,
      "grad_norm": 3.764255719469938,
      "learning_rate": 0.00015010779708205454,
      "loss": 1.0664,
      "step": 4897
    },
    {
      "epoch": 0.35305990052620195,
      "grad_norm": 0.38232011965664764,
      "learning_rate": 0.0001500875910419978,
      "loss": 1.0953,
      "step": 4898
    },
    {
      "epoch": 0.3531319829885389,
      "grad_norm": 0.1568175909163209,
      "learning_rate": 0.00015006738227171923,
      "loss": 1.0838,
      "step": 4899
    },
    {
      "epoch": 0.3532040654508758,
      "grad_norm": 0.23112225814202564,
      "learning_rate": 0.00015004717077232043,
      "loss": 1.0583,
      "step": 4900
    },
    {
      "epoch": 0.3532761479132127,
      "grad_norm": 0.32928426090533086,
      "learning_rate": 0.00015002695654490306,
      "loss": 1.0695,
      "step": 4901
    },
    {
      "epoch": 0.35334823037554963,
      "grad_norm": 0.21286649807818508,
      "learning_rate": 0.00015000673959056904,
      "loss": 1.1397,
      "step": 4902
    },
    {
      "epoch": 0.35342031283788655,
      "grad_norm": 0.13454168275356668,
      "learning_rate": 0.0001499865199104203,
      "loss": 1.0523,
      "step": 4903
    },
    {
      "epoch": 0.35349239530022347,
      "grad_norm": 0.4523543017496259,
      "learning_rate": 0.00014996629750555908,
      "loss": 1.0675,
      "step": 4904
    },
    {
      "epoch": 0.3535644777625604,
      "grad_norm": 0.14612010447106363,
      "learning_rate": 0.00014994607237708762,
      "loss": 1.0443,
      "step": 4905
    },
    {
      "epoch": 0.35363656022489726,
      "grad_norm": 0.29836050360800725,
      "learning_rate": 0.0001499258445261084,
      "loss": 1.0306,
      "step": 4906
    },
    {
      "epoch": 0.3537086426872342,
      "grad_norm": 0.2790427168422393,
      "learning_rate": 0.00014990561395372397,
      "loss": 1.135,
      "step": 4907
    },
    {
      "epoch": 0.3537807251495711,
      "grad_norm": 1.0403728477734309,
      "learning_rate": 0.00014988538066103713,
      "loss": 1.0743,
      "step": 4908
    },
    {
      "epoch": 0.353852807611908,
      "grad_norm": 0.13971348794053204,
      "learning_rate": 0.00014986514464915076,
      "loss": 1.0988,
      "step": 4909
    },
    {
      "epoch": 0.35392489007424494,
      "grad_norm": 0.1535094869101102,
      "learning_rate": 0.00014984490591916786,
      "loss": 1.081,
      "step": 4910
    },
    {
      "epoch": 0.35399697253658186,
      "grad_norm": 0.2040866014631761,
      "learning_rate": 0.00014982466447219171,
      "loss": 1.0763,
      "step": 4911
    },
    {
      "epoch": 0.3540690549989188,
      "grad_norm": 0.2960554215978407,
      "learning_rate": 0.00014980442030932558,
      "loss": 0.9722,
      "step": 4912
    },
    {
      "epoch": 0.3541411374612557,
      "grad_norm": 0.14953990505173625,
      "learning_rate": 0.000149784173431673,
      "loss": 1.1185,
      "step": 4913
    },
    {
      "epoch": 0.3542132199235926,
      "grad_norm": 0.2360269187468459,
      "learning_rate": 0.00014976392384033756,
      "loss": 1.1455,
      "step": 4914
    },
    {
      "epoch": 0.3542853023859295,
      "grad_norm": 0.1451355995386524,
      "learning_rate": 0.0001497436715364231,
      "loss": 1.0424,
      "step": 4915
    },
    {
      "epoch": 0.3543573848482664,
      "grad_norm": 0.30385603444117987,
      "learning_rate": 0.00014972341652103347,
      "loss": 1.1285,
      "step": 4916
    },
    {
      "epoch": 0.3544294673106033,
      "grad_norm": 0.22749279419351084,
      "learning_rate": 0.00014970315879527286,
      "loss": 1.0856,
      "step": 4917
    },
    {
      "epoch": 0.35450154977294024,
      "grad_norm": 0.18486537802306202,
      "learning_rate": 0.0001496828983602454,
      "loss": 0.9755,
      "step": 4918
    },
    {
      "epoch": 0.35457363223527716,
      "grad_norm": 0.13142028293137106,
      "learning_rate": 0.00014966263521705553,
      "loss": 1.1306,
      "step": 4919
    },
    {
      "epoch": 0.3546457146976141,
      "grad_norm": 0.11358062486642115,
      "learning_rate": 0.00014964236936680774,
      "loss": 1.0022,
      "step": 4920
    },
    {
      "epoch": 0.354717797159951,
      "grad_norm": 0.15455402437202606,
      "learning_rate": 0.0001496221008106067,
      "loss": 1.0224,
      "step": 4921
    },
    {
      "epoch": 0.3547898796222879,
      "grad_norm": 0.21446430124818464,
      "learning_rate": 0.0001496018295495572,
      "loss": 1.0573,
      "step": 4922
    },
    {
      "epoch": 0.35486196208462484,
      "grad_norm": 0.13138169808026362,
      "learning_rate": 0.0001495815555847643,
      "loss": 1.0867,
      "step": 4923
    },
    {
      "epoch": 0.3549340445469617,
      "grad_norm": 0.2929833381203659,
      "learning_rate": 0.000149561278917333,
      "loss": 1.0846,
      "step": 4924
    },
    {
      "epoch": 0.3550061270092986,
      "grad_norm": 0.1451013748320384,
      "learning_rate": 0.00014954099954836864,
      "loss": 1.0443,
      "step": 4925
    },
    {
      "epoch": 0.35507820947163554,
      "grad_norm": 0.1111599258619994,
      "learning_rate": 0.0001495207174789766,
      "loss": 1.0767,
      "step": 4926
    },
    {
      "epoch": 0.35515029193397246,
      "grad_norm": 0.1651300274079987,
      "learning_rate": 0.00014950043271026243,
      "loss": 1.086,
      "step": 4927
    },
    {
      "epoch": 0.3552223743963094,
      "grad_norm": 0.23482951810982985,
      "learning_rate": 0.00014948014524333183,
      "loss": 1.0514,
      "step": 4928
    },
    {
      "epoch": 0.3552944568586463,
      "grad_norm": 0.12582056068308914,
      "learning_rate": 0.00014945985507929064,
      "loss": 1.0698,
      "step": 4929
    },
    {
      "epoch": 0.3553665393209832,
      "grad_norm": 0.11307073517429093,
      "learning_rate": 0.00014943956221924486,
      "loss": 1.0303,
      "step": 4930
    },
    {
      "epoch": 0.35543862178332014,
      "grad_norm": 0.12776623943698553,
      "learning_rate": 0.00014941926666430067,
      "loss": 1.0834,
      "step": 4931
    },
    {
      "epoch": 0.355510704245657,
      "grad_norm": 0.8535039104055206,
      "learning_rate": 0.00014939896841556431,
      "loss": 1.0971,
      "step": 4932
    },
    {
      "epoch": 0.3555827867079939,
      "grad_norm": 2.0583810405964273,
      "learning_rate": 0.00014937866747414223,
      "loss": 0.9658,
      "step": 4933
    },
    {
      "epoch": 0.35565486917033085,
      "grad_norm": 0.1426030045491663,
      "learning_rate": 0.00014935836384114107,
      "loss": 1.0501,
      "step": 4934
    },
    {
      "epoch": 0.35572695163266776,
      "grad_norm": 0.24326581499570074,
      "learning_rate": 0.00014933805751766744,
      "loss": 1.1096,
      "step": 4935
    },
    {
      "epoch": 0.3557990340950047,
      "grad_norm": 0.11913701363956772,
      "learning_rate": 0.00014931774850482832,
      "loss": 1.1261,
      "step": 4936
    },
    {
      "epoch": 0.3558711165573416,
      "grad_norm": 0.1161435134794661,
      "learning_rate": 0.0001492974368037307,
      "loss": 1.0298,
      "step": 4937
    },
    {
      "epoch": 0.3559431990196785,
      "grad_norm": 0.12239918501527113,
      "learning_rate": 0.00014927712241548174,
      "loss": 1.0784,
      "step": 4938
    },
    {
      "epoch": 0.35601528148201544,
      "grad_norm": 0.13340020884553958,
      "learning_rate": 0.00014925680534118876,
      "loss": 1.0695,
      "step": 4939
    },
    {
      "epoch": 0.35608736394435236,
      "grad_norm": 0.13190678437952696,
      "learning_rate": 0.00014923648558195924,
      "loss": 1.13,
      "step": 4940
    },
    {
      "epoch": 0.35615944640668923,
      "grad_norm": 0.43721348179280417,
      "learning_rate": 0.00014921616313890072,
      "loss": 1.0397,
      "step": 4941
    },
    {
      "epoch": 0.35623152886902615,
      "grad_norm": 0.1924719471062853,
      "learning_rate": 0.00014919583801312108,
      "loss": 0.9942,
      "step": 4942
    },
    {
      "epoch": 0.35630361133136307,
      "grad_norm": 0.12427241980846442,
      "learning_rate": 0.0001491755102057281,
      "loss": 1.0746,
      "step": 4943
    },
    {
      "epoch": 0.3563756937937,
      "grad_norm": 0.1214918241519542,
      "learning_rate": 0.0001491551797178299,
      "loss": 0.9867,
      "step": 4944
    },
    {
      "epoch": 0.3564477762560369,
      "grad_norm": 0.11359065694337014,
      "learning_rate": 0.00014913484655053463,
      "loss": 1.0122,
      "step": 4945
    },
    {
      "epoch": 0.3565198587183738,
      "grad_norm": 0.10591292048150089,
      "learning_rate": 0.00014911451070495068,
      "loss": 1.0536,
      "step": 4946
    },
    {
      "epoch": 0.35659194118071075,
      "grad_norm": 0.13864395562568718,
      "learning_rate": 0.00014909417218218648,
      "loss": 1.0505,
      "step": 4947
    },
    {
      "epoch": 0.35666402364304767,
      "grad_norm": 0.13263090974155337,
      "learning_rate": 0.00014907383098335073,
      "loss": 1.0806,
      "step": 4948
    },
    {
      "epoch": 0.3567361061053846,
      "grad_norm": 0.19785721933094183,
      "learning_rate": 0.0001490534871095521,
      "loss": 1.1116,
      "step": 4949
    },
    {
      "epoch": 0.35680818856772145,
      "grad_norm": 0.13964373790272083,
      "learning_rate": 0.00014903314056189962,
      "loss": 1.01,
      "step": 4950
    },
    {
      "epoch": 0.35688027103005837,
      "grad_norm": 0.22760075251209524,
      "learning_rate": 0.00014901279134150233,
      "loss": 1.0314,
      "step": 4951
    },
    {
      "epoch": 0.3569523534923953,
      "grad_norm": 0.11951508089966499,
      "learning_rate": 0.00014899243944946938,
      "loss": 1.0352,
      "step": 4952
    },
    {
      "epoch": 0.3570244359547322,
      "grad_norm": 0.1098857863877429,
      "learning_rate": 0.0001489720848869102,
      "loss": 1.0422,
      "step": 4953
    },
    {
      "epoch": 0.35709651841706913,
      "grad_norm": 0.1209872265676998,
      "learning_rate": 0.0001489517276549343,
      "loss": 1.0731,
      "step": 4954
    },
    {
      "epoch": 0.35716860087940605,
      "grad_norm": 0.19109137191744138,
      "learning_rate": 0.0001489313677546513,
      "loss": 1.0792,
      "step": 4955
    },
    {
      "epoch": 0.35724068334174297,
      "grad_norm": 0.15317979972534115,
      "learning_rate": 0.00014891100518717098,
      "loss": 1.0704,
      "step": 4956
    },
    {
      "epoch": 0.3573127658040799,
      "grad_norm": 0.2516016007337052,
      "learning_rate": 0.00014889063995360334,
      "loss": 1.0344,
      "step": 4957
    },
    {
      "epoch": 0.35738484826641675,
      "grad_norm": 0.5098610270867557,
      "learning_rate": 0.00014887027205505842,
      "loss": 1.1552,
      "step": 4958
    },
    {
      "epoch": 0.3574569307287537,
      "grad_norm": 0.1092653759097124,
      "learning_rate": 0.00014884990149264645,
      "loss": 1.1335,
      "step": 4959
    },
    {
      "epoch": 0.3575290131910906,
      "grad_norm": 0.18404502338566367,
      "learning_rate": 0.00014882952826747784,
      "loss": 1.0376,
      "step": 4960
    },
    {
      "epoch": 0.3576010956534275,
      "grad_norm": 0.11720157783508166,
      "learning_rate": 0.00014880915238066312,
      "loss": 1.0763,
      "step": 4961
    },
    {
      "epoch": 0.35767317811576443,
      "grad_norm": 0.6571661481264366,
      "learning_rate": 0.00014878877383331292,
      "loss": 1.0508,
      "step": 4962
    },
    {
      "epoch": 0.35774526057810135,
      "grad_norm": 0.11715945752643735,
      "learning_rate": 0.00014876839262653805,
      "loss": 0.9999,
      "step": 4963
    },
    {
      "epoch": 0.3578173430404383,
      "grad_norm": 0.14257409400882975,
      "learning_rate": 0.00014874800876144953,
      "loss": 1.144,
      "step": 4964
    },
    {
      "epoch": 0.3578894255027752,
      "grad_norm": 0.11675407523327277,
      "learning_rate": 0.00014872762223915837,
      "loss": 1.0985,
      "step": 4965
    },
    {
      "epoch": 0.3579615079651121,
      "grad_norm": 0.13329552494925767,
      "learning_rate": 0.0001487072330607759,
      "loss": 1.0546,
      "step": 4966
    },
    {
      "epoch": 0.358033590427449,
      "grad_norm": 0.30141992434899434,
      "learning_rate": 0.00014868684122741344,
      "loss": 1.0675,
      "step": 4967
    },
    {
      "epoch": 0.3581056728897859,
      "grad_norm": 0.37213280232676704,
      "learning_rate": 0.00014866644674018265,
      "loss": 0.9915,
      "step": 4968
    },
    {
      "epoch": 0.3581777553521228,
      "grad_norm": 1.998582745218176,
      "learning_rate": 0.00014864604960019504,
      "loss": 1.0846,
      "step": 4969
    },
    {
      "epoch": 0.35824983781445974,
      "grad_norm": 0.11161989618919763,
      "learning_rate": 0.00014862564980856258,
      "loss": 1.153,
      "step": 4970
    },
    {
      "epoch": 0.35832192027679666,
      "grad_norm": 0.16646437917163853,
      "learning_rate": 0.00014860524736639717,
      "loss": 1.0586,
      "step": 4971
    },
    {
      "epoch": 0.3583940027391336,
      "grad_norm": 0.11015208665537457,
      "learning_rate": 0.00014858484227481096,
      "loss": 1.1034,
      "step": 4972
    },
    {
      "epoch": 0.3584660852014705,
      "grad_norm": 0.13302018055900888,
      "learning_rate": 0.00014856443453491617,
      "loss": 0.9568,
      "step": 4973
    },
    {
      "epoch": 0.3585381676638074,
      "grad_norm": 0.11653513508728305,
      "learning_rate": 0.00014854402414782525,
      "loss": 1.02,
      "step": 4974
    },
    {
      "epoch": 0.35861025012614434,
      "grad_norm": 0.14384916248689852,
      "learning_rate": 0.0001485236111146507,
      "loss": 1.0777,
      "step": 4975
    },
    {
      "epoch": 0.3586823325884812,
      "grad_norm": 0.14827655048590907,
      "learning_rate": 0.00014850319543650525,
      "loss": 1.0721,
      "step": 4976
    },
    {
      "epoch": 0.3587544150508181,
      "grad_norm": 0.13378649785343805,
      "learning_rate": 0.00014848277711450176,
      "loss": 1.0296,
      "step": 4977
    },
    {
      "epoch": 0.35882649751315504,
      "grad_norm": 0.15258504844075474,
      "learning_rate": 0.00014846235614975314,
      "loss": 1.072,
      "step": 4978
    },
    {
      "epoch": 0.35889857997549196,
      "grad_norm": 0.1679566096328321,
      "learning_rate": 0.00014844193254337257,
      "loss": 1.1418,
      "step": 4979
    },
    {
      "epoch": 0.3589706624378289,
      "grad_norm": 0.7578781933329812,
      "learning_rate": 0.0001484215062964733,
      "loss": 1.1482,
      "step": 4980
    },
    {
      "epoch": 0.3590427449001658,
      "grad_norm": 0.12803328894619107,
      "learning_rate": 0.00014840107741016873,
      "loss": 1.0462,
      "step": 4981
    },
    {
      "epoch": 0.3591148273625027,
      "grad_norm": 1.1567631359792148,
      "learning_rate": 0.00014838064588557247,
      "loss": 1.0034,
      "step": 4982
    },
    {
      "epoch": 0.35918690982483964,
      "grad_norm": 1.5478432604933423,
      "learning_rate": 0.00014836021172379816,
      "loss": 0.9834,
      "step": 4983
    },
    {
      "epoch": 0.3592589922871765,
      "grad_norm": 0.12475992072107026,
      "learning_rate": 0.00014833977492595968,
      "loss": 1.096,
      "step": 4984
    },
    {
      "epoch": 0.3593310747495134,
      "grad_norm": 0.12004224560177428,
      "learning_rate": 0.00014831933549317095,
      "loss": 1.1135,
      "step": 4985
    },
    {
      "epoch": 0.35940315721185034,
      "grad_norm": 0.13835505536484377,
      "learning_rate": 0.00014829889342654623,
      "loss": 1.1698,
      "step": 4986
    },
    {
      "epoch": 0.35947523967418726,
      "grad_norm": 0.10626361557574682,
      "learning_rate": 0.00014827844872719966,
      "loss": 1.0254,
      "step": 4987
    },
    {
      "epoch": 0.3595473221365242,
      "grad_norm": 0.19618356542730436,
      "learning_rate": 0.0001482580013962458,
      "loss": 1.0417,
      "step": 4988
    },
    {
      "epoch": 0.3596194045988611,
      "grad_norm": 0.6776509942560808,
      "learning_rate": 0.00014823755143479907,
      "loss": 1.0976,
      "step": 4989
    },
    {
      "epoch": 0.359691487061198,
      "grad_norm": 0.40994162876982404,
      "learning_rate": 0.00014821709884397429,
      "loss": 1.0967,
      "step": 4990
    },
    {
      "epoch": 0.35976356952353494,
      "grad_norm": 0.18385841322574473,
      "learning_rate": 0.0001481966436248862,
      "loss": 1.1065,
      "step": 4991
    },
    {
      "epoch": 0.35983565198587186,
      "grad_norm": 0.14346888072988329,
      "learning_rate": 0.00014817618577864992,
      "loss": 1.0726,
      "step": 4992
    },
    {
      "epoch": 0.3599077344482087,
      "grad_norm": 0.15584483594702023,
      "learning_rate": 0.00014815572530638046,
      "loss": 1.153,
      "step": 4993
    },
    {
      "epoch": 0.35997981691054565,
      "grad_norm": 0.20658067417748432,
      "learning_rate": 0.0001481352622091932,
      "loss": 1.0482,
      "step": 4994
    },
    {
      "epoch": 0.36005189937288257,
      "grad_norm": 0.18117988982852193,
      "learning_rate": 0.0001481147964882035,
      "loss": 1.0697,
      "step": 4995
    },
    {
      "epoch": 0.3601239818352195,
      "grad_norm": 0.12348490858758356,
      "learning_rate": 0.00014809432814452694,
      "loss": 1.0037,
      "step": 4996
    },
    {
      "epoch": 0.3601960642975564,
      "grad_norm": 0.21962092908552777,
      "learning_rate": 0.00014807385717927925,
      "loss": 1.1197,
      "step": 4997
    },
    {
      "epoch": 0.3602681467598933,
      "grad_norm": 0.11896600069300972,
      "learning_rate": 0.00014805338359357627,
      "loss": 1.0131,
      "step": 4998
    },
    {
      "epoch": 0.36034022922223025,
      "grad_norm": 0.18095157141854523,
      "learning_rate": 0.00014803290738853395,
      "loss": 1.0441,
      "step": 4999
    },
    {
      "epoch": 0.36041231168456717,
      "grad_norm": 0.1263051407704998,
      "learning_rate": 0.0001480124285652685,
      "loss": 1.0838,
      "step": 5000
    },
    {
      "epoch": 0.3604843941469041,
      "grad_norm": 0.1354813472033164,
      "learning_rate": 0.00014799194712489613,
      "loss": 0.9846,
      "step": 5001
    },
    {
      "epoch": 0.36055647660924095,
      "grad_norm": 0.12643248059062923,
      "learning_rate": 0.0001479714630685333,
      "loss": 1.0538,
      "step": 5002
    },
    {
      "epoch": 0.36062855907157787,
      "grad_norm": 0.10416574161165199,
      "learning_rate": 0.00014795097639729658,
      "loss": 1.0575,
      "step": 5003
    },
    {
      "epoch": 0.3607006415339148,
      "grad_norm": 0.12160078545340947,
      "learning_rate": 0.00014793048711230267,
      "loss": 1.0923,
      "step": 5004
    },
    {
      "epoch": 0.3607727239962517,
      "grad_norm": 0.12983700868505502,
      "learning_rate": 0.0001479099952146684,
      "loss": 1.147,
      "step": 5005
    },
    {
      "epoch": 0.36084480645858863,
      "grad_norm": 0.39292011341812105,
      "learning_rate": 0.00014788950070551078,
      "loss": 1.1938,
      "step": 5006
    },
    {
      "epoch": 0.36091688892092555,
      "grad_norm": 0.13038833515408793,
      "learning_rate": 0.00014786900358594693,
      "loss": 1.0859,
      "step": 5007
    },
    {
      "epoch": 0.36098897138326247,
      "grad_norm": 0.12430534355314465,
      "learning_rate": 0.00014784850385709413,
      "loss": 1.0866,
      "step": 5008
    },
    {
      "epoch": 0.3610610538455994,
      "grad_norm": 1.7843777080437615,
      "learning_rate": 0.00014782800152006984,
      "loss": 1.0453,
      "step": 5009
    },
    {
      "epoch": 0.36113313630793625,
      "grad_norm": 0.11274477102165609,
      "learning_rate": 0.00014780749657599155,
      "loss": 1.1042,
      "step": 5010
    },
    {
      "epoch": 0.3612052187702732,
      "grad_norm": 0.17880153187612402,
      "learning_rate": 0.00014778698902597704,
      "loss": 1.0305,
      "step": 5011
    },
    {
      "epoch": 0.3612773012326101,
      "grad_norm": 0.11830668304800537,
      "learning_rate": 0.00014776647887114407,
      "loss": 1.1296,
      "step": 5012
    },
    {
      "epoch": 0.361349383694947,
      "grad_norm": 0.1447752609936345,
      "learning_rate": 0.0001477459661126107,
      "loss": 1.0363,
      "step": 5013
    },
    {
      "epoch": 0.36142146615728393,
      "grad_norm": 0.1695426284422676,
      "learning_rate": 0.000147725450751495,
      "loss": 1.0105,
      "step": 5014
    },
    {
      "epoch": 0.36149354861962085,
      "grad_norm": 0.21244309316879442,
      "learning_rate": 0.00014770493278891528,
      "loss": 1.1035,
      "step": 5015
    },
    {
      "epoch": 0.3615656310819578,
      "grad_norm": 0.11393391923227973,
      "learning_rate": 0.00014768441222598997,
      "loss": 1.0747,
      "step": 5016
    },
    {
      "epoch": 0.3616377135442947,
      "grad_norm": 0.18123544746154205,
      "learning_rate": 0.00014766388906383758,
      "loss": 1.0487,
      "step": 5017
    },
    {
      "epoch": 0.3617097960066316,
      "grad_norm": 0.10869559837801496,
      "learning_rate": 0.0001476433633035768,
      "loss": 1.1173,
      "step": 5018
    },
    {
      "epoch": 0.3617818784689685,
      "grad_norm": 0.131450251650441,
      "learning_rate": 0.00014762283494632655,
      "loss": 1.103,
      "step": 5019
    },
    {
      "epoch": 0.3618539609313054,
      "grad_norm": 0.10933775231660044,
      "learning_rate": 0.00014760230399320568,
      "loss": 0.9854,
      "step": 5020
    },
    {
      "epoch": 0.3619260433936423,
      "grad_norm": 0.14198817309551687,
      "learning_rate": 0.0001475817704453334,
      "loss": 1.0309,
      "step": 5021
    },
    {
      "epoch": 0.36199812585597924,
      "grad_norm": 0.156064671368478,
      "learning_rate": 0.00014756123430382896,
      "loss": 1.0992,
      "step": 5022
    },
    {
      "epoch": 0.36207020831831616,
      "grad_norm": 0.11686512438498797,
      "learning_rate": 0.00014754069556981178,
      "loss": 1.0073,
      "step": 5023
    },
    {
      "epoch": 0.3621422907806531,
      "grad_norm": 0.1308890267008535,
      "learning_rate": 0.00014752015424440132,
      "loss": 1.0733,
      "step": 5024
    },
    {
      "epoch": 0.36221437324299,
      "grad_norm": 0.11560652186350928,
      "learning_rate": 0.0001474996103287174,
      "loss": 1.0324,
      "step": 5025
    },
    {
      "epoch": 0.3622864557053269,
      "grad_norm": 0.1275954546601504,
      "learning_rate": 0.0001474790638238797,
      "loss": 1.1301,
      "step": 5026
    },
    {
      "epoch": 0.36235853816766384,
      "grad_norm": 0.1081856287406518,
      "learning_rate": 0.00014745851473100833,
      "loss": 1.0473,
      "step": 5027
    },
    {
      "epoch": 0.3624306206300007,
      "grad_norm": 0.13633540102585343,
      "learning_rate": 0.00014743796305122331,
      "loss": 1.0514,
      "step": 5028
    },
    {
      "epoch": 0.3625027030923376,
      "grad_norm": 0.12947558891976507,
      "learning_rate": 0.0001474174087856449,
      "loss": 1.017,
      "step": 5029
    },
    {
      "epoch": 0.36257478555467454,
      "grad_norm": 0.1664915294616814,
      "learning_rate": 0.0001473968519353935,
      "loss": 1.0841,
      "step": 5030
    },
    {
      "epoch": 0.36264686801701146,
      "grad_norm": 0.11998009994043043,
      "learning_rate": 0.00014737629250158963,
      "loss": 0.9991,
      "step": 5031
    },
    {
      "epoch": 0.3627189504793484,
      "grad_norm": 0.10635877475318184,
      "learning_rate": 0.000147355730485354,
      "loss": 1.0401,
      "step": 5032
    },
    {
      "epoch": 0.3627910329416853,
      "grad_norm": 0.12150251895322266,
      "learning_rate": 0.0001473351658878074,
      "loss": 1.1273,
      "step": 5033
    },
    {
      "epoch": 0.3628631154040222,
      "grad_norm": 0.3095551334080991,
      "learning_rate": 0.00014731459871007078,
      "loss": 1.0312,
      "step": 5034
    },
    {
      "epoch": 0.36293519786635914,
      "grad_norm": 0.11117404646526394,
      "learning_rate": 0.00014729402895326526,
      "loss": 1.0942,
      "step": 5035
    },
    {
      "epoch": 0.363007280328696,
      "grad_norm": 0.12039813963804316,
      "learning_rate": 0.00014727345661851203,
      "loss": 0.9533,
      "step": 5036
    },
    {
      "epoch": 0.3630793627910329,
      "grad_norm": 0.12553263682932517,
      "learning_rate": 0.00014725288170693251,
      "loss": 1.0082,
      "step": 5037
    },
    {
      "epoch": 0.36315144525336984,
      "grad_norm": 0.11212468907103168,
      "learning_rate": 0.00014723230421964821,
      "loss": 1.0849,
      "step": 5038
    },
    {
      "epoch": 0.36322352771570676,
      "grad_norm": 0.12649600401655106,
      "learning_rate": 0.0001472117241577808,
      "loss": 1.0804,
      "step": 5039
    },
    {
      "epoch": 0.3632956101780437,
      "grad_norm": 0.14769294623167023,
      "learning_rate": 0.00014719114152245198,
      "loss": 1.1378,
      "step": 5040
    },
    {
      "epoch": 0.3633676926403806,
      "grad_norm": 0.1203093469077824,
      "learning_rate": 0.00014717055631478383,
      "loss": 1.0319,
      "step": 5041
    },
    {
      "epoch": 0.3634397751027175,
      "grad_norm": 0.18083562215867485,
      "learning_rate": 0.0001471499685358983,
      "loss": 1.0567,
      "step": 5042
    },
    {
      "epoch": 0.36351185756505444,
      "grad_norm": 0.23571841778326466,
      "learning_rate": 0.00014712937818691776,
      "loss": 1.1107,
      "step": 5043
    },
    {
      "epoch": 0.36358394002739136,
      "grad_norm": 0.19059325433575455,
      "learning_rate": 0.0001471087852689644,
      "loss": 1.0753,
      "step": 5044
    },
    {
      "epoch": 0.3636560224897282,
      "grad_norm": 0.12555706392272764,
      "learning_rate": 0.0001470881897831608,
      "loss": 0.9908,
      "step": 5045
    },
    {
      "epoch": 0.36372810495206515,
      "grad_norm": 0.11943374103547097,
      "learning_rate": 0.00014706759173062963,
      "loss": 1.0638,
      "step": 5046
    },
    {
      "epoch": 0.36380018741440207,
      "grad_norm": 2.079569850202351,
      "learning_rate": 0.00014704699111249363,
      "loss": 1.0981,
      "step": 5047
    },
    {
      "epoch": 0.363872269876739,
      "grad_norm": 0.11342002200766778,
      "learning_rate": 0.00014702638792987572,
      "loss": 1.0678,
      "step": 5048
    },
    {
      "epoch": 0.3639443523390759,
      "grad_norm": 0.10939880468205042,
      "learning_rate": 0.00014700578218389892,
      "loss": 1.0407,
      "step": 5049
    },
    {
      "epoch": 0.3640164348014128,
      "grad_norm": 0.10987433098461562,
      "learning_rate": 0.00014698517387568648,
      "loss": 1.1337,
      "step": 5050
    },
    {
      "epoch": 0.36408851726374974,
      "grad_norm": 0.1130231318398829,
      "learning_rate": 0.00014696456300636172,
      "loss": 1.112,
      "step": 5051
    },
    {
      "epoch": 0.36416059972608666,
      "grad_norm": 0.11330289198568462,
      "learning_rate": 0.00014694394957704814,
      "loss": 1.0529,
      "step": 5052
    },
    {
      "epoch": 0.36423268218842353,
      "grad_norm": 0.34859252496337245,
      "learning_rate": 0.00014692333358886934,
      "loss": 1.0376,
      "step": 5053
    },
    {
      "epoch": 0.36430476465076045,
      "grad_norm": 0.1279517159494143,
      "learning_rate": 0.00014690271504294903,
      "loss": 1.0131,
      "step": 5054
    },
    {
      "epoch": 0.36437684711309737,
      "grad_norm": 0.11638704795949617,
      "learning_rate": 0.0001468820939404112,
      "loss": 1.0062,
      "step": 5055
    },
    {
      "epoch": 0.3644489295754343,
      "grad_norm": 0.2011980278520308,
      "learning_rate": 0.0001468614702823798,
      "loss": 1.0178,
      "step": 5056
    },
    {
      "epoch": 0.3645210120377712,
      "grad_norm": 0.12209484225894385,
      "learning_rate": 0.00014684084406997903,
      "loss": 1.0405,
      "step": 5057
    },
    {
      "epoch": 0.36459309450010813,
      "grad_norm": 0.13721122247155834,
      "learning_rate": 0.0001468202153043332,
      "loss": 1.0527,
      "step": 5058
    },
    {
      "epoch": 0.36466517696244505,
      "grad_norm": 0.14097632005202937,
      "learning_rate": 0.00014679958398656682,
      "loss": 1.0386,
      "step": 5059
    },
    {
      "epoch": 0.36473725942478197,
      "grad_norm": 0.1224208610713215,
      "learning_rate": 0.0001467789501178044,
      "loss": 1.0773,
      "step": 5060
    },
    {
      "epoch": 0.3648093418871189,
      "grad_norm": 0.1302036283951279,
      "learning_rate": 0.00014675831369917068,
      "loss": 1.0721,
      "step": 5061
    },
    {
      "epoch": 0.36488142434945575,
      "grad_norm": 0.12104862206768763,
      "learning_rate": 0.00014673767473179057,
      "loss": 1.0992,
      "step": 5062
    },
    {
      "epoch": 0.36495350681179267,
      "grad_norm": 0.12027040463767992,
      "learning_rate": 0.00014671703321678906,
      "loss": 0.9751,
      "step": 5063
    },
    {
      "epoch": 0.3650255892741296,
      "grad_norm": 0.12557885003365213,
      "learning_rate": 0.0001466963891552913,
      "loss": 1.0701,
      "step": 5064
    },
    {
      "epoch": 0.3650976717364665,
      "grad_norm": 0.13357380327509755,
      "learning_rate": 0.00014667574254842257,
      "loss": 1.0452,
      "step": 5065
    },
    {
      "epoch": 0.36516975419880343,
      "grad_norm": 0.13742758043239328,
      "learning_rate": 0.0001466550933973083,
      "loss": 1.0681,
      "step": 5066
    },
    {
      "epoch": 0.36524183666114035,
      "grad_norm": 0.14715558137267534,
      "learning_rate": 0.00014663444170307405,
      "loss": 1.0175,
      "step": 5067
    },
    {
      "epoch": 0.36531391912347727,
      "grad_norm": 0.114103204318847,
      "learning_rate": 0.00014661378746684553,
      "loss": 1.0564,
      "step": 5068
    },
    {
      "epoch": 0.3653860015858142,
      "grad_norm": 0.12064099516031888,
      "learning_rate": 0.00014659313068974858,
      "loss": 1.1243,
      "step": 5069
    },
    {
      "epoch": 0.3654580840481511,
      "grad_norm": 0.10794630983886519,
      "learning_rate": 0.00014657247137290916,
      "loss": 1.0506,
      "step": 5070
    },
    {
      "epoch": 0.365530166510488,
      "grad_norm": 0.11902702965196402,
      "learning_rate": 0.0001465518095174534,
      "loss": 0.9749,
      "step": 5071
    },
    {
      "epoch": 0.3656022489728249,
      "grad_norm": 0.13323105860075504,
      "learning_rate": 0.00014653114512450756,
      "loss": 1.0499,
      "step": 5072
    },
    {
      "epoch": 0.3656743314351618,
      "grad_norm": 0.1131015799085786,
      "learning_rate": 0.00014651047819519804,
      "loss": 1.0778,
      "step": 5073
    },
    {
      "epoch": 0.36574641389749873,
      "grad_norm": 0.12264397608239248,
      "learning_rate": 0.00014648980873065135,
      "loss": 1.0402,
      "step": 5074
    },
    {
      "epoch": 0.36581849635983565,
      "grad_norm": 0.16701153086927506,
      "learning_rate": 0.00014646913673199417,
      "loss": 1.0953,
      "step": 5075
    },
    {
      "epoch": 0.3658905788221726,
      "grad_norm": 0.12986730340895855,
      "learning_rate": 0.00014644846220035333,
      "loss": 1.1024,
      "step": 5076
    },
    {
      "epoch": 0.3659626612845095,
      "grad_norm": 0.13183353411832796,
      "learning_rate": 0.00014642778513685576,
      "loss": 1.0711,
      "step": 5077
    },
    {
      "epoch": 0.3660347437468464,
      "grad_norm": 0.6071697892083587,
      "learning_rate": 0.00014640710554262852,
      "loss": 1.0353,
      "step": 5078
    },
    {
      "epoch": 0.3661068262091833,
      "grad_norm": 0.1508069921435764,
      "learning_rate": 0.0001463864234187989,
      "loss": 1.1187,
      "step": 5079
    },
    {
      "epoch": 0.3661789086715202,
      "grad_norm": 0.12060077761713042,
      "learning_rate": 0.00014636573876649419,
      "loss": 1.0232,
      "step": 5080
    },
    {
      "epoch": 0.3662509911338571,
      "grad_norm": 0.24229534660117777,
      "learning_rate": 0.0001463450515868419,
      "loss": 1.0601,
      "step": 5081
    },
    {
      "epoch": 0.36632307359619404,
      "grad_norm": 0.2763403847836795,
      "learning_rate": 0.00014632436188096972,
      "loss": 1.0511,
      "step": 5082
    },
    {
      "epoch": 0.36639515605853096,
      "grad_norm": 0.11500183062959884,
      "learning_rate": 0.0001463036696500054,
      "loss": 1.1133,
      "step": 5083
    },
    {
      "epoch": 0.3664672385208679,
      "grad_norm": 0.18495982259496496,
      "learning_rate": 0.0001462829748950768,
      "loss": 1.0521,
      "step": 5084
    },
    {
      "epoch": 0.3665393209832048,
      "grad_norm": 0.5242729216027036,
      "learning_rate": 0.000146262277617312,
      "loss": 1.0308,
      "step": 5085
    },
    {
      "epoch": 0.3666114034455417,
      "grad_norm": 0.12265189762262708,
      "learning_rate": 0.00014624157781783926,
      "loss": 1.084,
      "step": 5086
    },
    {
      "epoch": 0.36668348590787864,
      "grad_norm": 0.11894797716776077,
      "learning_rate": 0.00014622087549778678,
      "loss": 1.0562,
      "step": 5087
    },
    {
      "epoch": 0.3667555683702155,
      "grad_norm": 0.1157920936256049,
      "learning_rate": 0.00014620017065828308,
      "loss": 1.1148,
      "step": 5088
    },
    {
      "epoch": 0.3668276508325524,
      "grad_norm": 0.16383159273624912,
      "learning_rate": 0.0001461794633004568,
      "loss": 1.006,
      "step": 5089
    },
    {
      "epoch": 0.36689973329488934,
      "grad_norm": 0.29396926760491543,
      "learning_rate": 0.00014615875342543661,
      "loss": 1.0378,
      "step": 5090
    },
    {
      "epoch": 0.36697181575722626,
      "grad_norm": 0.11702794172647941,
      "learning_rate": 0.00014613804103435143,
      "loss": 1.0583,
      "step": 5091
    },
    {
      "epoch": 0.3670438982195632,
      "grad_norm": 0.11457078603569425,
      "learning_rate": 0.00014611732612833026,
      "loss": 1.1206,
      "step": 5092
    },
    {
      "epoch": 0.3671159806819001,
      "grad_norm": 0.11614911837566064,
      "learning_rate": 0.0001460966087085022,
      "loss": 1.0618,
      "step": 5093
    },
    {
      "epoch": 0.367188063144237,
      "grad_norm": 0.11364883192970593,
      "learning_rate": 0.0001460758887759966,
      "loss": 1.0202,
      "step": 5094
    },
    {
      "epoch": 0.36726014560657394,
      "grad_norm": 0.12196472179257807,
      "learning_rate": 0.00014605516633194282,
      "loss": 1.0484,
      "step": 5095
    },
    {
      "epoch": 0.36733222806891086,
      "grad_norm": 0.12928623046191892,
      "learning_rate": 0.0001460344413774705,
      "loss": 1.033,
      "step": 5096
    },
    {
      "epoch": 0.3674043105312477,
      "grad_norm": 0.17443879575513474,
      "learning_rate": 0.00014601371391370925,
      "loss": 0.9438,
      "step": 5097
    },
    {
      "epoch": 0.36747639299358464,
      "grad_norm": 0.1278801449807875,
      "learning_rate": 0.00014599298394178897,
      "loss": 1.0765,
      "step": 5098
    },
    {
      "epoch": 0.36754847545592156,
      "grad_norm": 0.11111629288715241,
      "learning_rate": 0.00014597225146283958,
      "loss": 1.0541,
      "step": 5099
    },
    {
      "epoch": 0.3676205579182585,
      "grad_norm": 0.15863092967130077,
      "learning_rate": 0.00014595151647799125,
      "loss": 1.0864,
      "step": 5100
    },
    {
      "epoch": 0.3676926403805954,
      "grad_norm": 0.11296821208107607,
      "learning_rate": 0.0001459307789883741,
      "loss": 1.0547,
      "step": 5101
    },
    {
      "epoch": 0.3677647228429323,
      "grad_norm": 0.11046840371860966,
      "learning_rate": 0.00014591003899511865,
      "loss": 1.1364,
      "step": 5102
    },
    {
      "epoch": 0.36783680530526924,
      "grad_norm": 0.11112025234710596,
      "learning_rate": 0.00014588929649935535,
      "loss": 1.0673,
      "step": 5103
    },
    {
      "epoch": 0.36790888776760616,
      "grad_norm": 0.12078077496684905,
      "learning_rate": 0.00014586855150221483,
      "loss": 1.017,
      "step": 5104
    },
    {
      "epoch": 0.367980970229943,
      "grad_norm": 0.14582983817930337,
      "learning_rate": 0.0001458478040048279,
      "loss": 0.9907,
      "step": 5105
    },
    {
      "epoch": 0.36805305269227995,
      "grad_norm": 0.1146809450438387,
      "learning_rate": 0.0001458270540083255,
      "loss": 1.0071,
      "step": 5106
    },
    {
      "epoch": 0.36812513515461687,
      "grad_norm": 0.14003497527714168,
      "learning_rate": 0.00014580630151383867,
      "loss": 1.0475,
      "step": 5107
    },
    {
      "epoch": 0.3681972176169538,
      "grad_norm": 0.11338775034499668,
      "learning_rate": 0.00014578554652249862,
      "loss": 1.0415,
      "step": 5108
    },
    {
      "epoch": 0.3682693000792907,
      "grad_norm": 0.14490694628523473,
      "learning_rate": 0.00014576478903543665,
      "loss": 1.0943,
      "step": 5109
    },
    {
      "epoch": 0.3683413825416276,
      "grad_norm": 0.11232604411668995,
      "learning_rate": 0.0001457440290537843,
      "loss": 1.0284,
      "step": 5110
    },
    {
      "epoch": 0.36841346500396455,
      "grad_norm": 0.1516234950931951,
      "learning_rate": 0.0001457232665786731,
      "loss": 1.0553,
      "step": 5111
    },
    {
      "epoch": 0.36848554746630147,
      "grad_norm": 0.13639265074121418,
      "learning_rate": 0.0001457025016112348,
      "loss": 1.04,
      "step": 5112
    },
    {
      "epoch": 0.3685576299286384,
      "grad_norm": 0.1868412689984815,
      "learning_rate": 0.0001456817341526013,
      "loss": 1.0597,
      "step": 5113
    },
    {
      "epoch": 0.36862971239097525,
      "grad_norm": 0.12480898411948593,
      "learning_rate": 0.00014566096420390464,
      "loss": 1.0626,
      "step": 5114
    },
    {
      "epoch": 0.36870179485331217,
      "grad_norm": 0.12139303309812384,
      "learning_rate": 0.0001456401917662769,
      "loss": 1.0483,
      "step": 5115
    },
    {
      "epoch": 0.3687738773156491,
      "grad_norm": 0.10766992270904258,
      "learning_rate": 0.00014561941684085045,
      "loss": 1.0548,
      "step": 5116
    },
    {
      "epoch": 0.368845959777986,
      "grad_norm": 0.10421133717577367,
      "learning_rate": 0.0001455986394287576,
      "loss": 1.0174,
      "step": 5117
    },
    {
      "epoch": 0.36891804224032293,
      "grad_norm": 0.11086757207894721,
      "learning_rate": 0.000145577859531131,
      "loss": 0.9511,
      "step": 5118
    },
    {
      "epoch": 0.36899012470265985,
      "grad_norm": 0.11318840360114964,
      "learning_rate": 0.00014555707714910328,
      "loss": 1.0238,
      "step": 5119
    },
    {
      "epoch": 0.36906220716499677,
      "grad_norm": 0.14134036683771736,
      "learning_rate": 0.00014553629228380733,
      "loss": 1.0584,
      "step": 5120
    },
    {
      "epoch": 0.3691342896273337,
      "grad_norm": 0.11566245664358486,
      "learning_rate": 0.00014551550493637606,
      "loss": 1.0253,
      "step": 5121
    },
    {
      "epoch": 0.3692063720896706,
      "grad_norm": 0.17761314568170866,
      "learning_rate": 0.00014549471510794256,
      "loss": 1.0923,
      "step": 5122
    },
    {
      "epoch": 0.3692784545520075,
      "grad_norm": 0.11729558996014905,
      "learning_rate": 0.00014547392279964008,
      "loss": 1.0271,
      "step": 5123
    },
    {
      "epoch": 0.3693505370143444,
      "grad_norm": 0.14463657981698336,
      "learning_rate": 0.00014545312801260202,
      "loss": 1.0186,
      "step": 5124
    },
    {
      "epoch": 0.3694226194766813,
      "grad_norm": 0.12365236676479575,
      "learning_rate": 0.0001454323307479618,
      "loss": 0.9997,
      "step": 5125
    },
    {
      "epoch": 0.36949470193901823,
      "grad_norm": 0.11426258906206242,
      "learning_rate": 0.00014541153100685312,
      "loss": 1.0907,
      "step": 5126
    },
    {
      "epoch": 0.36956678440135515,
      "grad_norm": 0.11423143337345751,
      "learning_rate": 0.00014539072879040976,
      "loss": 1.0147,
      "step": 5127
    },
    {
      "epoch": 0.3696388668636921,
      "grad_norm": 0.15432957477333611,
      "learning_rate": 0.00014536992409976558,
      "loss": 1.0068,
      "step": 5128
    },
    {
      "epoch": 0.369710949326029,
      "grad_norm": 0.13794066940554908,
      "learning_rate": 0.00014534911693605464,
      "loss": 1.0892,
      "step": 5129
    },
    {
      "epoch": 0.3697830317883659,
      "grad_norm": 0.3049026191508098,
      "learning_rate": 0.00014532830730041113,
      "loss": 1.1015,
      "step": 5130
    },
    {
      "epoch": 0.3698551142507028,
      "grad_norm": 0.1848292316569893,
      "learning_rate": 0.00014530749519396936,
      "loss": 1.0249,
      "step": 5131
    },
    {
      "epoch": 0.3699271967130397,
      "grad_norm": 0.6657344370630037,
      "learning_rate": 0.00014528668061786378,
      "loss": 1.0692,
      "step": 5132
    },
    {
      "epoch": 0.3699992791753766,
      "grad_norm": 0.12242782663579557,
      "learning_rate": 0.00014526586357322893,
      "loss": 1.1201,
      "step": 5133
    },
    {
      "epoch": 0.37007136163771354,
      "grad_norm": 0.6672117634553492,
      "learning_rate": 0.00014524504406119958,
      "loss": 1.1271,
      "step": 5134
    },
    {
      "epoch": 0.37014344410005046,
      "grad_norm": 0.6558977297615781,
      "learning_rate": 0.00014522422208291052,
      "loss": 1.1175,
      "step": 5135
    },
    {
      "epoch": 0.3702155265623874,
      "grad_norm": 0.13798302756197328,
      "learning_rate": 0.00014520339763949682,
      "loss": 1.0207,
      "step": 5136
    },
    {
      "epoch": 0.3702876090247243,
      "grad_norm": 0.19381382777516976,
      "learning_rate": 0.00014518257073209352,
      "loss": 1.0357,
      "step": 5137
    },
    {
      "epoch": 0.3703596914870612,
      "grad_norm": 0.12490035177276351,
      "learning_rate": 0.00014516174136183592,
      "loss": 1.049,
      "step": 5138
    },
    {
      "epoch": 0.37043177394939814,
      "grad_norm": 0.1256239874150437,
      "learning_rate": 0.00014514090952985934,
      "loss": 1.0738,
      "step": 5139
    },
    {
      "epoch": 0.370503856411735,
      "grad_norm": 0.1574850700736399,
      "learning_rate": 0.0001451200752372994,
      "loss": 1.0788,
      "step": 5140
    },
    {
      "epoch": 0.3705759388740719,
      "grad_norm": 0.1053395961385796,
      "learning_rate": 0.00014509923848529164,
      "loss": 1.009,
      "step": 5141
    },
    {
      "epoch": 0.37064802133640884,
      "grad_norm": 0.11710489690357337,
      "learning_rate": 0.000145078399274972,
      "loss": 1.0595,
      "step": 5142
    },
    {
      "epoch": 0.37072010379874576,
      "grad_norm": 0.2740948484547928,
      "learning_rate": 0.00014505755760747626,
      "loss": 1.0969,
      "step": 5143
    },
    {
      "epoch": 0.3707921862610827,
      "grad_norm": 0.10754139852129671,
      "learning_rate": 0.00014503671348394057,
      "loss": 1.0997,
      "step": 5144
    },
    {
      "epoch": 0.3708642687234196,
      "grad_norm": 0.1108476068829066,
      "learning_rate": 0.00014501586690550102,
      "loss": 0.9962,
      "step": 5145
    },
    {
      "epoch": 0.3709363511857565,
      "grad_norm": 0.11209351714282838,
      "learning_rate": 0.00014499501787329408,
      "loss": 1.0316,
      "step": 5146
    },
    {
      "epoch": 0.37100843364809344,
      "grad_norm": 0.41909717118185574,
      "learning_rate": 0.0001449741663884561,
      "loss": 1.0869,
      "step": 5147
    },
    {
      "epoch": 0.37108051611043036,
      "grad_norm": 0.11794306225788673,
      "learning_rate": 0.00014495331245212366,
      "loss": 1.0478,
      "step": 5148
    },
    {
      "epoch": 0.3711525985727672,
      "grad_norm": 0.1343368209690954,
      "learning_rate": 0.0001449324560654336,
      "loss": 1.066,
      "step": 5149
    },
    {
      "epoch": 0.37122468103510414,
      "grad_norm": 0.43846398482214793,
      "learning_rate": 0.00014491159722952265,
      "loss": 1.1394,
      "step": 5150
    },
    {
      "epoch": 0.37129676349744106,
      "grad_norm": 0.24329806371967233,
      "learning_rate": 0.00014489073594552786,
      "loss": 1.0678,
      "step": 5151
    },
    {
      "epoch": 0.371368845959778,
      "grad_norm": 0.24435172897651,
      "learning_rate": 0.0001448698722145864,
      "loss": 1.0464,
      "step": 5152
    },
    {
      "epoch": 0.3714409284221149,
      "grad_norm": 0.29541540104871916,
      "learning_rate": 0.00014484900603783544,
      "loss": 1.1106,
      "step": 5153
    },
    {
      "epoch": 0.3715130108844518,
      "grad_norm": 0.14336597834994233,
      "learning_rate": 0.00014482813741641243,
      "loss": 0.998,
      "step": 5154
    },
    {
      "epoch": 0.37158509334678874,
      "grad_norm": 0.3088658708957395,
      "learning_rate": 0.00014480726635145491,
      "loss": 1.0193,
      "step": 5155
    },
    {
      "epoch": 0.37165717580912566,
      "grad_norm": 0.14719200448080835,
      "learning_rate": 0.0001447863928441005,
      "loss": 1.0163,
      "step": 5156
    },
    {
      "epoch": 0.3717292582714625,
      "grad_norm": 0.12205890465081477,
      "learning_rate": 0.00014476551689548704,
      "loss": 1.0787,
      "step": 5157
    },
    {
      "epoch": 0.37180134073379945,
      "grad_norm": 0.11798714291547002,
      "learning_rate": 0.0001447446385067524,
      "loss": 1.0637,
      "step": 5158
    },
    {
      "epoch": 0.37187342319613637,
      "grad_norm": 0.12521191493661118,
      "learning_rate": 0.00014472375767903464,
      "loss": 1.0289,
      "step": 5159
    },
    {
      "epoch": 0.3719455056584733,
      "grad_norm": 0.10808463273283635,
      "learning_rate": 0.00014470287441347202,
      "loss": 1.0527,
      "step": 5160
    },
    {
      "epoch": 0.3720175881208102,
      "grad_norm": 0.11726314237876502,
      "learning_rate": 0.0001446819887112028,
      "loss": 1.1327,
      "step": 5161
    },
    {
      "epoch": 0.3720896705831471,
      "grad_norm": 0.11834303943027187,
      "learning_rate": 0.00014466110057336545,
      "loss": 1.0388,
      "step": 5162
    },
    {
      "epoch": 0.37216175304548405,
      "grad_norm": 0.13350456116576226,
      "learning_rate": 0.00014464021000109857,
      "loss": 1.0327,
      "step": 5163
    },
    {
      "epoch": 0.37223383550782096,
      "grad_norm": 0.1635472308075495,
      "learning_rate": 0.0001446193169955409,
      "loss": 1.1155,
      "step": 5164
    },
    {
      "epoch": 0.3723059179701579,
      "grad_norm": 0.14820128441420843,
      "learning_rate": 0.00014459842155783127,
      "loss": 1.0376,
      "step": 5165
    },
    {
      "epoch": 0.37237800043249475,
      "grad_norm": 0.14403951459612568,
      "learning_rate": 0.00014457752368910866,
      "loss": 1.1407,
      "step": 5166
    },
    {
      "epoch": 0.37245008289483167,
      "grad_norm": 0.11858383354303585,
      "learning_rate": 0.00014455662339051223,
      "loss": 1.0405,
      "step": 5167
    },
    {
      "epoch": 0.3725221653571686,
      "grad_norm": 0.12840298984399168,
      "learning_rate": 0.00014453572066318123,
      "loss": 1.0582,
      "step": 5168
    },
    {
      "epoch": 0.3725942478195055,
      "grad_norm": 0.127804192110216,
      "learning_rate": 0.00014451481550825502,
      "loss": 1.0589,
      "step": 5169
    },
    {
      "epoch": 0.37266633028184243,
      "grad_norm": 0.18820000813594723,
      "learning_rate": 0.00014449390792687305,
      "loss": 1.0786,
      "step": 5170
    },
    {
      "epoch": 0.37273841274417935,
      "grad_norm": 0.1263523608855955,
      "learning_rate": 0.00014447299792017513,
      "loss": 1.0599,
      "step": 5171
    },
    {
      "epoch": 0.37281049520651627,
      "grad_norm": 0.25654634338910626,
      "learning_rate": 0.0001444520854893009,
      "loss": 1.0971,
      "step": 5172
    },
    {
      "epoch": 0.3728825776688532,
      "grad_norm": 0.16896940600490343,
      "learning_rate": 0.00014443117063539038,
      "loss": 1.059,
      "step": 5173
    },
    {
      "epoch": 0.3729546601311901,
      "grad_norm": 0.11359025988036864,
      "learning_rate": 0.0001444102533595835,
      "loss": 1.0996,
      "step": 5174
    },
    {
      "epoch": 0.37302674259352697,
      "grad_norm": 0.13698864838095856,
      "learning_rate": 0.00014438933366302058,
      "loss": 1.0058,
      "step": 5175
    },
    {
      "epoch": 0.3730988250558639,
      "grad_norm": 0.11125378135850672,
      "learning_rate": 0.00014436841154684182,
      "loss": 1.044,
      "step": 5176
    },
    {
      "epoch": 0.3731709075182008,
      "grad_norm": 0.11639730527807583,
      "learning_rate": 0.00014434748701218773,
      "loss": 1.0331,
      "step": 5177
    },
    {
      "epoch": 0.37324298998053773,
      "grad_norm": 0.2900584653455006,
      "learning_rate": 0.0001443265600601988,
      "loss": 1.091,
      "step": 5178
    },
    {
      "epoch": 0.37331507244287465,
      "grad_norm": 0.12314316143197981,
      "learning_rate": 0.00014430563069201585,
      "loss": 0.9856,
      "step": 5179
    },
    {
      "epoch": 0.37338715490521157,
      "grad_norm": 0.12892441646410602,
      "learning_rate": 0.0001442846989087796,
      "loss": 1.0206,
      "step": 5180
    },
    {
      "epoch": 0.3734592373675485,
      "grad_norm": 0.15417411695350036,
      "learning_rate": 0.00014426376471163112,
      "loss": 1.0826,
      "step": 5181
    },
    {
      "epoch": 0.3735313198298854,
      "grad_norm": 0.1431206437254308,
      "learning_rate": 0.00014424282810171146,
      "loss": 1.0201,
      "step": 5182
    },
    {
      "epoch": 0.3736034022922223,
      "grad_norm": 0.11460213538305918,
      "learning_rate": 0.00014422188908016183,
      "loss": 1.0773,
      "step": 5183
    },
    {
      "epoch": 0.3736754847545592,
      "grad_norm": 0.14744066469901093,
      "learning_rate": 0.00014420094764812366,
      "loss": 1.0328,
      "step": 5184
    },
    {
      "epoch": 0.3737475672168961,
      "grad_norm": 0.1045441452653548,
      "learning_rate": 0.0001441800038067384,
      "loss": 1.0593,
      "step": 5185
    },
    {
      "epoch": 0.37381964967923303,
      "grad_norm": 0.8704206988607851,
      "learning_rate": 0.0001441590575571477,
      "loss": 1.0632,
      "step": 5186
    },
    {
      "epoch": 0.37389173214156995,
      "grad_norm": 0.13427569766234934,
      "learning_rate": 0.00014413810890049326,
      "loss": 1.0333,
      "step": 5187
    },
    {
      "epoch": 0.3739638146039069,
      "grad_norm": 0.19833054396384117,
      "learning_rate": 0.00014411715783791706,
      "loss": 1.1051,
      "step": 5188
    },
    {
      "epoch": 0.3740358970662438,
      "grad_norm": 0.11378591672241035,
      "learning_rate": 0.00014409620437056105,
      "loss": 1.0457,
      "step": 5189
    },
    {
      "epoch": 0.3741079795285807,
      "grad_norm": 0.12279042627537987,
      "learning_rate": 0.0001440752484995674,
      "loss": 1.1064,
      "step": 5190
    },
    {
      "epoch": 0.37418006199091763,
      "grad_norm": 0.11101634038426594,
      "learning_rate": 0.00014405429022607845,
      "loss": 1.0281,
      "step": 5191
    },
    {
      "epoch": 0.3742521444532545,
      "grad_norm": 0.11468236689044307,
      "learning_rate": 0.00014403332955123651,
      "loss": 1.0608,
      "step": 5192
    },
    {
      "epoch": 0.3743242269155914,
      "grad_norm": 0.13889935121990904,
      "learning_rate": 0.0001440123664761842,
      "loss": 1.0092,
      "step": 5193
    },
    {
      "epoch": 0.37439630937792834,
      "grad_norm": 0.13969232514141944,
      "learning_rate": 0.00014399140100206415,
      "loss": 1.0847,
      "step": 5194
    },
    {
      "epoch": 0.37446839184026526,
      "grad_norm": 0.12267164679809743,
      "learning_rate": 0.0001439704331300192,
      "loss": 1.0912,
      "step": 5195
    },
    {
      "epoch": 0.3745404743026022,
      "grad_norm": 0.12146448315229082,
      "learning_rate": 0.00014394946286119226,
      "loss": 1.1205,
      "step": 5196
    },
    {
      "epoch": 0.3746125567649391,
      "grad_norm": 0.14494040830980204,
      "learning_rate": 0.00014392849019672645,
      "loss": 1.0907,
      "step": 5197
    },
    {
      "epoch": 0.374684639227276,
      "grad_norm": 0.11757538985051531,
      "learning_rate": 0.00014390751513776489,
      "loss": 1.1187,
      "step": 5198
    },
    {
      "epoch": 0.37475672168961294,
      "grad_norm": 0.1352987077425531,
      "learning_rate": 0.00014388653768545098,
      "loss": 1.0299,
      "step": 5199
    },
    {
      "epoch": 0.37482880415194986,
      "grad_norm": 0.1478074617096099,
      "learning_rate": 0.0001438655578409281,
      "loss": 1.0256,
      "step": 5200
    },
    {
      "epoch": 0.3749008866142867,
      "grad_norm": 0.18964151776974666,
      "learning_rate": 0.00014384457560533994,
      "loss": 0.985,
      "step": 5201
    },
    {
      "epoch": 0.37497296907662364,
      "grad_norm": 1.403293876282266,
      "learning_rate": 0.00014382359097983013,
      "loss": 1.067,
      "step": 5202
    },
    {
      "epoch": 0.37504505153896056,
      "grad_norm": 0.11990982423566857,
      "learning_rate": 0.00014380260396554254,
      "loss": 1.1054,
      "step": 5203
    },
    {
      "epoch": 0.3751171340012975,
      "grad_norm": 0.13831117278361235,
      "learning_rate": 0.0001437816145636212,
      "loss": 1.0085,
      "step": 5204
    },
    {
      "epoch": 0.3751892164636344,
      "grad_norm": 0.13771490292112273,
      "learning_rate": 0.00014376062277521016,
      "loss": 1.0498,
      "step": 5205
    },
    {
      "epoch": 0.3752612989259713,
      "grad_norm": 0.13299187638903096,
      "learning_rate": 0.0001437396286014537,
      "loss": 1.0446,
      "step": 5206
    },
    {
      "epoch": 0.37533338138830824,
      "grad_norm": 0.13632844270095656,
      "learning_rate": 0.0001437186320434962,
      "loss": 1.0331,
      "step": 5207
    },
    {
      "epoch": 0.37540546385064516,
      "grad_norm": 0.1314799247740535,
      "learning_rate": 0.00014369763310248206,
      "loss": 1.1115,
      "step": 5208
    },
    {
      "epoch": 0.375477546312982,
      "grad_norm": 0.14354300720731236,
      "learning_rate": 0.00014367663177955605,
      "loss": 1.1398,
      "step": 5209
    },
    {
      "epoch": 0.37554962877531894,
      "grad_norm": 0.1660242727502551,
      "learning_rate": 0.00014365562807586282,
      "loss": 1.0875,
      "step": 5210
    },
    {
      "epoch": 0.37562171123765586,
      "grad_norm": 0.13402308101476149,
      "learning_rate": 0.00014363462199254735,
      "loss": 1.0047,
      "step": 5211
    },
    {
      "epoch": 0.3756937936999928,
      "grad_norm": 0.11342021924076948,
      "learning_rate": 0.0001436136135307546,
      "loss": 1.0339,
      "step": 5212
    },
    {
      "epoch": 0.3757658761623297,
      "grad_norm": 0.46264182262614684,
      "learning_rate": 0.00014359260269162976,
      "loss": 1.1019,
      "step": 5213
    },
    {
      "epoch": 0.3758379586246666,
      "grad_norm": 0.3004102765323605,
      "learning_rate": 0.00014357158947631802,
      "loss": 0.9862,
      "step": 5214
    },
    {
      "epoch": 0.37591004108700354,
      "grad_norm": 0.12114427642956045,
      "learning_rate": 0.00014355057388596492,
      "loss": 1.1208,
      "step": 5215
    },
    {
      "epoch": 0.37598212354934046,
      "grad_norm": 0.16403761320540158,
      "learning_rate": 0.00014352955592171588,
      "loss": 1.012,
      "step": 5216
    },
    {
      "epoch": 0.3760542060116774,
      "grad_norm": 0.15669200410612108,
      "learning_rate": 0.00014350853558471664,
      "loss": 1.0075,
      "step": 5217
    },
    {
      "epoch": 0.37612628847401425,
      "grad_norm": 0.13533726771835142,
      "learning_rate": 0.00014348751287611292,
      "loss": 1.0245,
      "step": 5218
    },
    {
      "epoch": 0.37619837093635117,
      "grad_norm": 0.16105093321901845,
      "learning_rate": 0.00014346648779705077,
      "loss": 1.0468,
      "step": 5219
    },
    {
      "epoch": 0.3762704533986881,
      "grad_norm": 0.10680842517383365,
      "learning_rate": 0.0001434454603486761,
      "loss": 1.0821,
      "step": 5220
    },
    {
      "epoch": 0.376342535861025,
      "grad_norm": 0.10797351173595512,
      "learning_rate": 0.00014342443053213527,
      "loss": 1.0257,
      "step": 5221
    },
    {
      "epoch": 0.3764146183233619,
      "grad_norm": 0.3885914875787842,
      "learning_rate": 0.0001434033983485744,
      "loss": 1.0883,
      "step": 5222
    },
    {
      "epoch": 0.37648670078569885,
      "grad_norm": 0.11584680799670227,
      "learning_rate": 0.00014338236379914004,
      "loss": 1.0764,
      "step": 5223
    },
    {
      "epoch": 0.37655878324803577,
      "grad_norm": 0.14290787945918454,
      "learning_rate": 0.00014336132688497877,
      "loss": 1.1272,
      "step": 5224
    },
    {
      "epoch": 0.3766308657103727,
      "grad_norm": 0.16775731040690997,
      "learning_rate": 0.00014334028760723725,
      "loss": 1.1366,
      "step": 5225
    },
    {
      "epoch": 0.3767029481727096,
      "grad_norm": 0.12444343342922501,
      "learning_rate": 0.00014331924596706233,
      "loss": 1.0562,
      "step": 5226
    },
    {
      "epoch": 0.37677503063504647,
      "grad_norm": 0.13768191980710517,
      "learning_rate": 0.00014329820196560094,
      "loss": 1.065,
      "step": 5227
    },
    {
      "epoch": 0.3768471130973834,
      "grad_norm": 0.10899579356152249,
      "learning_rate": 0.0001432771556040002,
      "loss": 1.0334,
      "step": 5228
    },
    {
      "epoch": 0.3769191955597203,
      "grad_norm": 0.11974963390104387,
      "learning_rate": 0.0001432561068834073,
      "loss": 1.0664,
      "step": 5229
    },
    {
      "epoch": 0.37699127802205723,
      "grad_norm": 0.21306776577443992,
      "learning_rate": 0.0001432350558049696,
      "loss": 1.0817,
      "step": 5230
    },
    {
      "epoch": 0.37706336048439415,
      "grad_norm": 0.13081434452242396,
      "learning_rate": 0.00014321400236983457,
      "loss": 1.0229,
      "step": 5231
    },
    {
      "epoch": 0.37713544294673107,
      "grad_norm": 0.11147665379924503,
      "learning_rate": 0.0001431929465791498,
      "loss": 1.0904,
      "step": 5232
    },
    {
      "epoch": 0.377207525409068,
      "grad_norm": 0.1336623742247752,
      "learning_rate": 0.00014317188843406304,
      "loss": 0.9988,
      "step": 5233
    },
    {
      "epoch": 0.3772796078714049,
      "grad_norm": 0.13962573237664727,
      "learning_rate": 0.00014315082793572213,
      "loss": 1.0003,
      "step": 5234
    },
    {
      "epoch": 0.3773516903337418,
      "grad_norm": 0.17536443456319675,
      "learning_rate": 0.00014312976508527506,
      "loss": 1.0242,
      "step": 5235
    },
    {
      "epoch": 0.3774237727960787,
      "grad_norm": 0.1394759294363317,
      "learning_rate": 0.00014310869988386993,
      "loss": 1.1271,
      "step": 5236
    },
    {
      "epoch": 0.3774958552584156,
      "grad_norm": 0.22560271501297538,
      "learning_rate": 0.000143087632332655,
      "loss": 1.036,
      "step": 5237
    },
    {
      "epoch": 0.37756793772075253,
      "grad_norm": 0.16094136200429704,
      "learning_rate": 0.00014306656243277862,
      "loss": 1.0414,
      "step": 5238
    },
    {
      "epoch": 0.37764002018308945,
      "grad_norm": 0.2499020825873942,
      "learning_rate": 0.00014304549018538933,
      "loss": 0.9684,
      "step": 5239
    },
    {
      "epoch": 0.3777121026454264,
      "grad_norm": 0.12695773132927357,
      "learning_rate": 0.00014302441559163574,
      "loss": 1.0599,
      "step": 5240
    },
    {
      "epoch": 0.3777841851077633,
      "grad_norm": 0.10358993752633067,
      "learning_rate": 0.00014300333865266656,
      "loss": 1.103,
      "step": 5241
    },
    {
      "epoch": 0.3778562675701002,
      "grad_norm": 0.1536931463455552,
      "learning_rate": 0.0001429822593696307,
      "loss": 1.074,
      "step": 5242
    },
    {
      "epoch": 0.37792835003243713,
      "grad_norm": 0.10796948504846746,
      "learning_rate": 0.0001429611777436772,
      "loss": 1.0442,
      "step": 5243
    },
    {
      "epoch": 0.378000432494774,
      "grad_norm": 0.4331696349089261,
      "learning_rate": 0.00014294009377595516,
      "loss": 1.0791,
      "step": 5244
    },
    {
      "epoch": 0.3780725149571109,
      "grad_norm": 0.1020582574595184,
      "learning_rate": 0.00014291900746761385,
      "loss": 1.0528,
      "step": 5245
    },
    {
      "epoch": 0.37814459741944784,
      "grad_norm": 0.16183493496139698,
      "learning_rate": 0.0001428979188198027,
      "loss": 1.0549,
      "step": 5246
    },
    {
      "epoch": 0.37821667988178476,
      "grad_norm": 0.1136133875641103,
      "learning_rate": 0.00014287682783367117,
      "loss": 1.0857,
      "step": 5247
    },
    {
      "epoch": 0.3782887623441217,
      "grad_norm": 0.11584356543814664,
      "learning_rate": 0.00014285573451036892,
      "loss": 1.0581,
      "step": 5248
    },
    {
      "epoch": 0.3783608448064586,
      "grad_norm": 0.12703264310999726,
      "learning_rate": 0.00014283463885104576,
      "loss": 1.0839,
      "step": 5249
    },
    {
      "epoch": 0.3784329272687955,
      "grad_norm": 0.14905629861109992,
      "learning_rate": 0.00014281354085685162,
      "loss": 1.0423,
      "step": 5250
    },
    {
      "epoch": 0.37850500973113244,
      "grad_norm": 0.15485463599637408,
      "learning_rate": 0.0001427924405289364,
      "loss": 1.1013,
      "step": 5251
    },
    {
      "epoch": 0.37857709219346936,
      "grad_norm": 0.1370624848605864,
      "learning_rate": 0.0001427713378684504,
      "loss": 0.9909,
      "step": 5252
    },
    {
      "epoch": 0.3786491746558062,
      "grad_norm": 0.11687593565673796,
      "learning_rate": 0.0001427502328765438,
      "loss": 1.0354,
      "step": 5253
    },
    {
      "epoch": 0.37872125711814314,
      "grad_norm": 1.9371324557198615,
      "learning_rate": 0.00014272912555436708,
      "loss": 1.1106,
      "step": 5254
    },
    {
      "epoch": 0.37879333958048006,
      "grad_norm": 0.13295362967496857,
      "learning_rate": 0.00014270801590307073,
      "loss": 0.9965,
      "step": 5255
    },
    {
      "epoch": 0.378865422042817,
      "grad_norm": 0.10736261433962284,
      "learning_rate": 0.00014268690392380546,
      "loss": 0.9913,
      "step": 5256
    },
    {
      "epoch": 0.3789375045051539,
      "grad_norm": 0.12019581890937271,
      "learning_rate": 0.00014266578961772203,
      "loss": 1.0597,
      "step": 5257
    },
    {
      "epoch": 0.3790095869674908,
      "grad_norm": 0.11658164406874781,
      "learning_rate": 0.00014264467298597138,
      "loss": 1.1646,
      "step": 5258
    },
    {
      "epoch": 0.37908166942982774,
      "grad_norm": 0.144156163709119,
      "learning_rate": 0.0001426235540297045,
      "loss": 1.0397,
      "step": 5259
    },
    {
      "epoch": 0.37915375189216466,
      "grad_norm": 0.13505091328331661,
      "learning_rate": 0.00014260243275007265,
      "loss": 1.0718,
      "step": 5260
    },
    {
      "epoch": 0.3792258343545015,
      "grad_norm": 0.1591711545787405,
      "learning_rate": 0.00014258130914822706,
      "loss": 1.132,
      "step": 5261
    },
    {
      "epoch": 0.37929791681683844,
      "grad_norm": 0.16610891762906788,
      "learning_rate": 0.00014256018322531918,
      "loss": 1.1667,
      "step": 5262
    },
    {
      "epoch": 0.37936999927917536,
      "grad_norm": 0.11988350273982269,
      "learning_rate": 0.00014253905498250058,
      "loss": 0.9441,
      "step": 5263
    },
    {
      "epoch": 0.3794420817415123,
      "grad_norm": 0.1281340951159216,
      "learning_rate": 0.00014251792442092288,
      "loss": 1.08,
      "step": 5264
    },
    {
      "epoch": 0.3795141642038492,
      "grad_norm": 0.16255960488230284,
      "learning_rate": 0.00014249679154173796,
      "loss": 1.0412,
      "step": 5265
    },
    {
      "epoch": 0.3795862466661861,
      "grad_norm": 0.10659902996454526,
      "learning_rate": 0.0001424756563460977,
      "loss": 1.0113,
      "step": 5266
    },
    {
      "epoch": 0.37965832912852304,
      "grad_norm": 0.11949796474563415,
      "learning_rate": 0.00014245451883515418,
      "loss": 1.0313,
      "step": 5267
    },
    {
      "epoch": 0.37973041159085996,
      "grad_norm": 0.12509659698094724,
      "learning_rate": 0.00014243337901005958,
      "loss": 1.1037,
      "step": 5268
    },
    {
      "epoch": 0.3798024940531969,
      "grad_norm": 0.11327548188003612,
      "learning_rate": 0.00014241223687196615,
      "loss": 1.0261,
      "step": 5269
    },
    {
      "epoch": 0.37987457651553375,
      "grad_norm": 0.12328486125596597,
      "learning_rate": 0.00014239109242202645,
      "loss": 1.052,
      "step": 5270
    },
    {
      "epoch": 0.37994665897787067,
      "grad_norm": 0.10480605669536404,
      "learning_rate": 0.0001423699456613929,
      "loss": 1.0355,
      "step": 5271
    },
    {
      "epoch": 0.3800187414402076,
      "grad_norm": 0.12789292363963317,
      "learning_rate": 0.00014234879659121835,
      "loss": 1.0767,
      "step": 5272
    },
    {
      "epoch": 0.3800908239025445,
      "grad_norm": 0.130363314499737,
      "learning_rate": 0.00014232764521265542,
      "loss": 1.0467,
      "step": 5273
    },
    {
      "epoch": 0.3801629063648814,
      "grad_norm": 0.10636895857091315,
      "learning_rate": 0.00014230649152685722,
      "loss": 1.1073,
      "step": 5274
    },
    {
      "epoch": 0.38023498882721835,
      "grad_norm": 0.1506526296236099,
      "learning_rate": 0.00014228533553497673,
      "loss": 1.0994,
      "step": 5275
    },
    {
      "epoch": 0.38030707128955527,
      "grad_norm": 0.11374064421446574,
      "learning_rate": 0.00014226417723816717,
      "loss": 1.1035,
      "step": 5276
    },
    {
      "epoch": 0.3803791537518922,
      "grad_norm": 0.11763041099744509,
      "learning_rate": 0.00014224301663758182,
      "loss": 1.0892,
      "step": 5277
    },
    {
      "epoch": 0.3804512362142291,
      "grad_norm": 0.12619440967008724,
      "learning_rate": 0.00014222185373437415,
      "loss": 1.0526,
      "step": 5278
    },
    {
      "epoch": 0.38052331867656597,
      "grad_norm": 0.11390537269148991,
      "learning_rate": 0.00014220068852969773,
      "loss": 1.0158,
      "step": 5279
    },
    {
      "epoch": 0.3805954011389029,
      "grad_norm": 0.1283346957045519,
      "learning_rate": 0.00014217952102470627,
      "loss": 1.1704,
      "step": 5280
    },
    {
      "epoch": 0.3806674836012398,
      "grad_norm": 1.3453890120474277,
      "learning_rate": 0.00014215835122055354,
      "loss": 1.1188,
      "step": 5281
    },
    {
      "epoch": 0.38073956606357673,
      "grad_norm": 0.10850934629362732,
      "learning_rate": 0.0001421371791183935,
      "loss": 1.048,
      "step": 5282
    },
    {
      "epoch": 0.38081164852591365,
      "grad_norm": 0.11638660991225414,
      "learning_rate": 0.00014211600471938026,
      "loss": 1.0684,
      "step": 5283
    },
    {
      "epoch": 0.38088373098825057,
      "grad_norm": 0.12230913682464635,
      "learning_rate": 0.00014209482802466795,
      "loss": 1.0477,
      "step": 5284
    },
    {
      "epoch": 0.3809558134505875,
      "grad_norm": 0.23646785081464564,
      "learning_rate": 0.00014207364903541094,
      "loss": 0.9941,
      "step": 5285
    },
    {
      "epoch": 0.3810278959129244,
      "grad_norm": 0.12357989087975293,
      "learning_rate": 0.00014205246775276367,
      "loss": 1.0513,
      "step": 5286
    },
    {
      "epoch": 0.3810999783752613,
      "grad_norm": 0.11284343896606332,
      "learning_rate": 0.00014203128417788067,
      "loss": 1.0725,
      "step": 5287
    },
    {
      "epoch": 0.3811720608375982,
      "grad_norm": 0.14060999021257425,
      "learning_rate": 0.00014201009831191666,
      "loss": 1.0466,
      "step": 5288
    },
    {
      "epoch": 0.3812441432999351,
      "grad_norm": 0.12611668333178325,
      "learning_rate": 0.00014198891015602646,
      "loss": 1.0132,
      "step": 5289
    },
    {
      "epoch": 0.38131622576227203,
      "grad_norm": 0.11135928308782932,
      "learning_rate": 0.00014196771971136505,
      "loss": 1.016,
      "step": 5290
    },
    {
      "epoch": 0.38138830822460895,
      "grad_norm": 0.16288570378284714,
      "learning_rate": 0.0001419465269790874,
      "loss": 1.0532,
      "step": 5291
    },
    {
      "epoch": 0.38146039068694587,
      "grad_norm": 0.1892035638358673,
      "learning_rate": 0.0001419253319603488,
      "loss": 1.0941,
      "step": 5292
    },
    {
      "epoch": 0.3815324731492828,
      "grad_norm": 0.1324485252056817,
      "learning_rate": 0.00014190413465630451,
      "loss": 1.0887,
      "step": 5293
    },
    {
      "epoch": 0.3816045556116197,
      "grad_norm": 0.20197345859099208,
      "learning_rate": 0.00014188293506811007,
      "loss": 1.0132,
      "step": 5294
    },
    {
      "epoch": 0.38167663807395663,
      "grad_norm": 0.12860549877061503,
      "learning_rate": 0.0001418617331969209,
      "loss": 1.0485,
      "step": 5295
    },
    {
      "epoch": 0.3817487205362935,
      "grad_norm": 0.12083898894330598,
      "learning_rate": 0.00014184052904389277,
      "loss": 1.05,
      "step": 5296
    },
    {
      "epoch": 0.3818208029986304,
      "grad_norm": 0.11718048894408747,
      "learning_rate": 0.0001418193226101815,
      "loss": 1.1145,
      "step": 5297
    },
    {
      "epoch": 0.38189288546096734,
      "grad_norm": 0.13152725473926358,
      "learning_rate": 0.00014179811389694304,
      "loss": 1.1038,
      "step": 5298
    },
    {
      "epoch": 0.38196496792330425,
      "grad_norm": 0.2509171010150382,
      "learning_rate": 0.0001417769029053334,
      "loss": 1.0936,
      "step": 5299
    },
    {
      "epoch": 0.3820370503856412,
      "grad_norm": 0.11637868249322074,
      "learning_rate": 0.00014175568963650884,
      "loss": 1.0444,
      "step": 5300
    },
    {
      "epoch": 0.3821091328479781,
      "grad_norm": 0.1051120761424661,
      "learning_rate": 0.00014173447409162563,
      "loss": 1.0828,
      "step": 5301
    },
    {
      "epoch": 0.382181215310315,
      "grad_norm": 0.10001680302290428,
      "learning_rate": 0.00014171325627184022,
      "loss": 1.0463,
      "step": 5302
    },
    {
      "epoch": 0.38225329777265193,
      "grad_norm": 0.11334255613678892,
      "learning_rate": 0.00014169203617830914,
      "loss": 1.1117,
      "step": 5303
    },
    {
      "epoch": 0.38232538023498885,
      "grad_norm": 0.5155591248876021,
      "learning_rate": 0.00014167081381218912,
      "loss": 1.0425,
      "step": 5304
    },
    {
      "epoch": 0.3823974626973257,
      "grad_norm": 0.13501365569093035,
      "learning_rate": 0.00014164958917463696,
      "loss": 1.0425,
      "step": 5305
    },
    {
      "epoch": 0.38246954515966264,
      "grad_norm": 0.14012332394869617,
      "learning_rate": 0.00014162836226680955,
      "loss": 1.0766,
      "step": 5306
    },
    {
      "epoch": 0.38254162762199956,
      "grad_norm": 0.15462282826040424,
      "learning_rate": 0.000141607133089864,
      "loss": 1.0649,
      "step": 5307
    },
    {
      "epoch": 0.3826137100843365,
      "grad_norm": 0.260126941268993,
      "learning_rate": 0.00014158590164495746,
      "loss": 0.9997,
      "step": 5308
    },
    {
      "epoch": 0.3826857925466734,
      "grad_norm": 0.5196207492883538,
      "learning_rate": 0.00014156466793324726,
      "loss": 1.0266,
      "step": 5309
    },
    {
      "epoch": 0.3827578750090103,
      "grad_norm": 0.1753069157013582,
      "learning_rate": 0.00014154343195589078,
      "loss": 1.0579,
      "step": 5310
    },
    {
      "epoch": 0.38282995747134724,
      "grad_norm": 0.1892581651088856,
      "learning_rate": 0.00014152219371404564,
      "loss": 0.9623,
      "step": 5311
    },
    {
      "epoch": 0.38290203993368416,
      "grad_norm": 0.12833874268114065,
      "learning_rate": 0.00014150095320886942,
      "loss": 1.1431,
      "step": 5312
    },
    {
      "epoch": 0.382974122396021,
      "grad_norm": 0.1436177623072491,
      "learning_rate": 0.00014147971044152003,
      "loss": 1.0992,
      "step": 5313
    },
    {
      "epoch": 0.38304620485835794,
      "grad_norm": 0.8187423984094245,
      "learning_rate": 0.0001414584654131553,
      "loss": 1.0507,
      "step": 5314
    },
    {
      "epoch": 0.38311828732069486,
      "grad_norm": 0.14504263296791872,
      "learning_rate": 0.00014143721812493333,
      "loss": 1.1281,
      "step": 5315
    },
    {
      "epoch": 0.3831903697830318,
      "grad_norm": 0.10164724994806267,
      "learning_rate": 0.00014141596857801222,
      "loss": 0.9519,
      "step": 5316
    },
    {
      "epoch": 0.3832624522453687,
      "grad_norm": 0.15767493731254614,
      "learning_rate": 0.00014139471677355033,
      "loss": 1.0497,
      "step": 5317
    },
    {
      "epoch": 0.3833345347077056,
      "grad_norm": 0.14286193079236992,
      "learning_rate": 0.00014137346271270604,
      "loss": 1.1161,
      "step": 5318
    },
    {
      "epoch": 0.38340661717004254,
      "grad_norm": 0.13560740152390421,
      "learning_rate": 0.0001413522063966379,
      "loss": 1.0483,
      "step": 5319
    },
    {
      "epoch": 0.38347869963237946,
      "grad_norm": 0.12509960120455166,
      "learning_rate": 0.00014133094782650457,
      "loss": 1.002,
      "step": 5320
    },
    {
      "epoch": 0.3835507820947164,
      "grad_norm": 0.2040019526375948,
      "learning_rate": 0.00014130968700346483,
      "loss": 1.1285,
      "step": 5321
    },
    {
      "epoch": 0.38362286455705324,
      "grad_norm": 0.12627488773248985,
      "learning_rate": 0.00014128842392867758,
      "loss": 1.1316,
      "step": 5322
    },
    {
      "epoch": 0.38369494701939016,
      "grad_norm": 0.12668332368553392,
      "learning_rate": 0.00014126715860330187,
      "loss": 1.0249,
      "step": 5323
    },
    {
      "epoch": 0.3837670294817271,
      "grad_norm": 0.27688869729812376,
      "learning_rate": 0.00014124589102849677,
      "loss": 1.0417,
      "step": 5324
    },
    {
      "epoch": 0.383839111944064,
      "grad_norm": 0.12010427247377338,
      "learning_rate": 0.00014122462120542167,
      "loss": 1.0533,
      "step": 5325
    },
    {
      "epoch": 0.3839111944064009,
      "grad_norm": 0.3900876499254545,
      "learning_rate": 0.00014120334913523587,
      "loss": 1.1107,
      "step": 5326
    },
    {
      "epoch": 0.38398327686873784,
      "grad_norm": 0.15052484226029159,
      "learning_rate": 0.00014118207481909895,
      "loss": 1.0993,
      "step": 5327
    },
    {
      "epoch": 0.38405535933107476,
      "grad_norm": 0.10541101815283677,
      "learning_rate": 0.0001411607982581705,
      "loss": 1.0022,
      "step": 5328
    },
    {
      "epoch": 0.3841274417934117,
      "grad_norm": 0.16494369421098262,
      "learning_rate": 0.00014113951945361034,
      "loss": 1.061,
      "step": 5329
    },
    {
      "epoch": 0.3841995242557486,
      "grad_norm": 0.39772948816823545,
      "learning_rate": 0.0001411182384065783,
      "loss": 1.0529,
      "step": 5330
    },
    {
      "epoch": 0.38427160671808547,
      "grad_norm": 0.139831764835365,
      "learning_rate": 0.00014109695511823445,
      "loss": 0.9919,
      "step": 5331
    },
    {
      "epoch": 0.3843436891804224,
      "grad_norm": 0.21146518354562047,
      "learning_rate": 0.00014107566958973885,
      "loss": 1.1055,
      "step": 5332
    },
    {
      "epoch": 0.3844157716427593,
      "grad_norm": 0.1431220381597134,
      "learning_rate": 0.0001410543818222518,
      "loss": 1.106,
      "step": 5333
    },
    {
      "epoch": 0.3844878541050962,
      "grad_norm": 0.13641868631443582,
      "learning_rate": 0.00014103309181693366,
      "loss": 1.0971,
      "step": 5334
    },
    {
      "epoch": 0.38455993656743315,
      "grad_norm": 0.10844492094695199,
      "learning_rate": 0.0001410117995749449,
      "loss": 1.1017,
      "step": 5335
    },
    {
      "epoch": 0.38463201902977007,
      "grad_norm": 0.12343214681667572,
      "learning_rate": 0.00014099050509744617,
      "loss": 1.0799,
      "step": 5336
    },
    {
      "epoch": 0.384704101492107,
      "grad_norm": 0.1355945273959926,
      "learning_rate": 0.0001409692083855982,
      "loss": 1.0383,
      "step": 5337
    },
    {
      "epoch": 0.3847761839544439,
      "grad_norm": 0.26485889278677843,
      "learning_rate": 0.00014094790944056186,
      "loss": 1.0864,
      "step": 5338
    },
    {
      "epoch": 0.38484826641678077,
      "grad_norm": 0.13101478407049727,
      "learning_rate": 0.00014092660826349812,
      "loss": 1.0389,
      "step": 5339
    },
    {
      "epoch": 0.3849203488791177,
      "grad_norm": 0.12780351450526764,
      "learning_rate": 0.0001409053048555681,
      "loss": 1.0972,
      "step": 5340
    },
    {
      "epoch": 0.3849924313414546,
      "grad_norm": 0.2717615038797322,
      "learning_rate": 0.00014088399921793302,
      "loss": 1.0443,
      "step": 5341
    },
    {
      "epoch": 0.38506451380379153,
      "grad_norm": 0.12020054166252253,
      "learning_rate": 0.0001408626913517542,
      "loss": 1.0804,
      "step": 5342
    },
    {
      "epoch": 0.38513659626612845,
      "grad_norm": 0.13231020060050175,
      "learning_rate": 0.00014084138125819318,
      "loss": 1.1303,
      "step": 5343
    },
    {
      "epoch": 0.38520867872846537,
      "grad_norm": 0.12665457572559224,
      "learning_rate": 0.00014082006893841145,
      "loss": 1.0367,
      "step": 5344
    },
    {
      "epoch": 0.3852807611908023,
      "grad_norm": 0.11174133586533869,
      "learning_rate": 0.0001407987543935708,
      "loss": 1.0655,
      "step": 5345
    },
    {
      "epoch": 0.3853528436531392,
      "grad_norm": 0.1032848780004234,
      "learning_rate": 0.00014077743762483302,
      "loss": 0.9916,
      "step": 5346
    },
    {
      "epoch": 0.38542492611547613,
      "grad_norm": 0.11367974371669068,
      "learning_rate": 0.0001407561186333601,
      "loss": 0.9696,
      "step": 5347
    },
    {
      "epoch": 0.385497008577813,
      "grad_norm": 0.154414660147713,
      "learning_rate": 0.0001407347974203141,
      "loss": 1.0069,
      "step": 5348
    },
    {
      "epoch": 0.3855690910401499,
      "grad_norm": 0.15472838214859502,
      "learning_rate": 0.00014071347398685724,
      "loss": 1.0318,
      "step": 5349
    },
    {
      "epoch": 0.38564117350248683,
      "grad_norm": 0.12212709812908563,
      "learning_rate": 0.00014069214833415175,
      "loss": 1.0656,
      "step": 5350
    },
    {
      "epoch": 0.38571325596482375,
      "grad_norm": 0.12845700714085256,
      "learning_rate": 0.0001406708204633602,
      "loss": 1.1007,
      "step": 5351
    },
    {
      "epoch": 0.3857853384271607,
      "grad_norm": 0.1127953666364492,
      "learning_rate": 0.00014064949037564503,
      "loss": 0.9702,
      "step": 5352
    },
    {
      "epoch": 0.3858574208894976,
      "grad_norm": 0.11846882973126359,
      "learning_rate": 0.00014062815807216902,
      "loss": 1.0243,
      "step": 5353
    },
    {
      "epoch": 0.3859295033518345,
      "grad_norm": 0.13336463342635904,
      "learning_rate": 0.00014060682355409492,
      "loss": 1.0451,
      "step": 5354
    },
    {
      "epoch": 0.38600158581417143,
      "grad_norm": 0.12973950748365667,
      "learning_rate": 0.00014058548682258564,
      "loss": 1.037,
      "step": 5355
    },
    {
      "epoch": 0.38607366827650835,
      "grad_norm": 0.12417984067053096,
      "learning_rate": 0.00014056414787880423,
      "loss": 0.9865,
      "step": 5356
    },
    {
      "epoch": 0.3861457507388452,
      "grad_norm": 0.13578961145907129,
      "learning_rate": 0.0001405428067239139,
      "loss": 1.0843,
      "step": 5357
    },
    {
      "epoch": 0.38621783320118214,
      "grad_norm": 0.11638834930652088,
      "learning_rate": 0.00014052146335907786,
      "loss": 1.0272,
      "step": 5358
    },
    {
      "epoch": 0.38628991566351906,
      "grad_norm": 0.30714742177226056,
      "learning_rate": 0.00014050011778545955,
      "loss": 1.1107,
      "step": 5359
    },
    {
      "epoch": 0.386361998125856,
      "grad_norm": 0.14559603464983326,
      "learning_rate": 0.00014047877000422253,
      "loss": 1.103,
      "step": 5360
    },
    {
      "epoch": 0.3864340805881929,
      "grad_norm": 0.23001772078432722,
      "learning_rate": 0.00014045742001653037,
      "loss": 1.0441,
      "step": 5361
    },
    {
      "epoch": 0.3865061630505298,
      "grad_norm": 0.11082530401692159,
      "learning_rate": 0.00014043606782354693,
      "loss": 1.115,
      "step": 5362
    },
    {
      "epoch": 0.38657824551286674,
      "grad_norm": 1.1423198674654025,
      "learning_rate": 0.00014041471342643598,
      "loss": 1.0882,
      "step": 5363
    },
    {
      "epoch": 0.38665032797520366,
      "grad_norm": 1.8365485592386994,
      "learning_rate": 0.00014039335682636164,
      "loss": 1.128,
      "step": 5364
    },
    {
      "epoch": 0.3867224104375405,
      "grad_norm": 0.11282541908910759,
      "learning_rate": 0.00014037199802448798,
      "loss": 1.1027,
      "step": 5365
    },
    {
      "epoch": 0.38679449289987744,
      "grad_norm": 0.19648735928311165,
      "learning_rate": 0.00014035063702197922,
      "loss": 1.0329,
      "step": 5366
    },
    {
      "epoch": 0.38686657536221436,
      "grad_norm": 0.1678467217762039,
      "learning_rate": 0.00014032927381999976,
      "loss": 1.0746,
      "step": 5367
    },
    {
      "epoch": 0.3869386578245513,
      "grad_norm": 0.22486656785099784,
      "learning_rate": 0.0001403079084197141,
      "loss": 1.0093,
      "step": 5368
    },
    {
      "epoch": 0.3870107402868882,
      "grad_norm": 0.13136943998182626,
      "learning_rate": 0.00014028654082228682,
      "loss": 1.021,
      "step": 5369
    },
    {
      "epoch": 0.3870828227492251,
      "grad_norm": 0.14788576177225085,
      "learning_rate": 0.0001402651710288826,
      "loss": 1.1204,
      "step": 5370
    },
    {
      "epoch": 0.38715490521156204,
      "grad_norm": 0.11149082621589244,
      "learning_rate": 0.00014024379904066642,
      "loss": 1.0917,
      "step": 5371
    },
    {
      "epoch": 0.38722698767389896,
      "grad_norm": 0.10819101895447505,
      "learning_rate": 0.0001402224248588031,
      "loss": 1.1028,
      "step": 5372
    },
    {
      "epoch": 0.3872990701362359,
      "grad_norm": 0.1204052209599107,
      "learning_rate": 0.00014020104848445783,
      "loss": 1.1714,
      "step": 5373
    },
    {
      "epoch": 0.38737115259857274,
      "grad_norm": 0.13613466793863158,
      "learning_rate": 0.00014017966991879574,
      "loss": 1.0604,
      "step": 5374
    },
    {
      "epoch": 0.38744323506090966,
      "grad_norm": 0.12607801289453646,
      "learning_rate": 0.0001401582891629822,
      "loss": 0.9909,
      "step": 5375
    },
    {
      "epoch": 0.3875153175232466,
      "grad_norm": 0.116153443759843,
      "learning_rate": 0.00014013690621818262,
      "loss": 1.0425,
      "step": 5376
    },
    {
      "epoch": 0.3875873999855835,
      "grad_norm": 0.1160721505784813,
      "learning_rate": 0.0001401155210855626,
      "loss": 1.0826,
      "step": 5377
    },
    {
      "epoch": 0.3876594824479204,
      "grad_norm": 0.1617376465513817,
      "learning_rate": 0.00014009413376628776,
      "loss": 1.0368,
      "step": 5378
    },
    {
      "epoch": 0.38773156491025734,
      "grad_norm": 0.14617711861841093,
      "learning_rate": 0.00014007274426152395,
      "loss": 1.0194,
      "step": 5379
    },
    {
      "epoch": 0.38780364737259426,
      "grad_norm": 0.6881847516444475,
      "learning_rate": 0.0001400513525724371,
      "loss": 1.0607,
      "step": 5380
    },
    {
      "epoch": 0.3878757298349312,
      "grad_norm": 0.12013691330992432,
      "learning_rate": 0.0001400299587001932,
      "loss": 1.0681,
      "step": 5381
    },
    {
      "epoch": 0.3879478122972681,
      "grad_norm": 0.17195823567785068,
      "learning_rate": 0.00014000856264595842,
      "loss": 1.0544,
      "step": 5382
    },
    {
      "epoch": 0.38801989475960497,
      "grad_norm": 0.14474106009123283,
      "learning_rate": 0.00013998716441089907,
      "loss": 1.0093,
      "step": 5383
    },
    {
      "epoch": 0.3880919772219419,
      "grad_norm": 0.11780973570018394,
      "learning_rate": 0.00013996576399618154,
      "loss": 1.0435,
      "step": 5384
    },
    {
      "epoch": 0.3881640596842788,
      "grad_norm": 0.11905480568182314,
      "learning_rate": 0.0001399443614029723,
      "loss": 1.0822,
      "step": 5385
    },
    {
      "epoch": 0.3882361421466157,
      "grad_norm": 0.12986635344952224,
      "learning_rate": 0.000139922956632438,
      "loss": 1.0515,
      "step": 5386
    },
    {
      "epoch": 0.38830822460895265,
      "grad_norm": 0.36543730728107543,
      "learning_rate": 0.00013990154968574544,
      "loss": 1.1362,
      "step": 5387
    },
    {
      "epoch": 0.38838030707128957,
      "grad_norm": 0.12294284716272594,
      "learning_rate": 0.00013988014056406144,
      "loss": 1.085,
      "step": 5388
    },
    {
      "epoch": 0.3884523895336265,
      "grad_norm": 0.11916903747848823,
      "learning_rate": 0.000139858729268553,
      "loss": 1.104,
      "step": 5389
    },
    {
      "epoch": 0.3885244719959634,
      "grad_norm": 0.1335743344295318,
      "learning_rate": 0.00013983731580038722,
      "loss": 0.9886,
      "step": 5390
    },
    {
      "epoch": 0.38859655445830027,
      "grad_norm": 0.19262415985685838,
      "learning_rate": 0.00013981590016073133,
      "loss": 1.0983,
      "step": 5391
    },
    {
      "epoch": 0.3886686369206372,
      "grad_norm": 0.13075780452452845,
      "learning_rate": 0.0001397944823507527,
      "loss": 1.1103,
      "step": 5392
    },
    {
      "epoch": 0.3887407193829741,
      "grad_norm": 0.1750125437263303,
      "learning_rate": 0.00013977306237161877,
      "loss": 1.001,
      "step": 5393
    },
    {
      "epoch": 0.38881280184531103,
      "grad_norm": 0.2147182115247554,
      "learning_rate": 0.00013975164022449709,
      "loss": 1.0916,
      "step": 5394
    },
    {
      "epoch": 0.38888488430764795,
      "grad_norm": 0.14837361712260627,
      "learning_rate": 0.0001397302159105554,
      "loss": 1.0213,
      "step": 5395
    },
    {
      "epoch": 0.38895696676998487,
      "grad_norm": 0.11853742966427942,
      "learning_rate": 0.00013970878943096152,
      "loss": 1.0407,
      "step": 5396
    },
    {
      "epoch": 0.3890290492323218,
      "grad_norm": 0.1553449249504325,
      "learning_rate": 0.00013968736078688338,
      "loss": 1.0931,
      "step": 5397
    },
    {
      "epoch": 0.3891011316946587,
      "grad_norm": 0.12523841692961665,
      "learning_rate": 0.00013966592997948903,
      "loss": 1.0182,
      "step": 5398
    },
    {
      "epoch": 0.38917321415699563,
      "grad_norm": 0.14872122062539864,
      "learning_rate": 0.0001396444970099466,
      "loss": 0.9997,
      "step": 5399
    },
    {
      "epoch": 0.3892452966193325,
      "grad_norm": 0.12331254564389992,
      "learning_rate": 0.00013962306187942445,
      "loss": 1.052,
      "step": 5400
    },
    {
      "epoch": 0.3893173790816694,
      "grad_norm": 0.09612280728521669,
      "learning_rate": 0.0001396016245890909,
      "loss": 1.0655,
      "step": 5401
    },
    {
      "epoch": 0.38938946154400633,
      "grad_norm": 0.11568137912446137,
      "learning_rate": 0.00013958018514011457,
      "loss": 1.0379,
      "step": 5402
    },
    {
      "epoch": 0.38946154400634325,
      "grad_norm": 0.1130245889075194,
      "learning_rate": 0.000139558743533664,
      "loss": 1.0181,
      "step": 5403
    },
    {
      "epoch": 0.38953362646868017,
      "grad_norm": 0.25766395329703795,
      "learning_rate": 0.0001395372997709081,
      "loss": 1.1564,
      "step": 5404
    },
    {
      "epoch": 0.3896057089310171,
      "grad_norm": 0.10624813779257994,
      "learning_rate": 0.00013951585385301555,
      "loss": 1.0634,
      "step": 5405
    },
    {
      "epoch": 0.389677791393354,
      "grad_norm": 0.11398590414633518,
      "learning_rate": 0.0001394944057811555,
      "loss": 1.0307,
      "step": 5406
    },
    {
      "epoch": 0.38974987385569093,
      "grad_norm": 0.10638143986014172,
      "learning_rate": 0.000139472955556497,
      "loss": 1.0246,
      "step": 5407
    },
    {
      "epoch": 0.38982195631802785,
      "grad_norm": 0.11806777331862486,
      "learning_rate": 0.00013945150318020927,
      "loss": 1.0951,
      "step": 5408
    },
    {
      "epoch": 0.3898940387803647,
      "grad_norm": 0.14515662998015694,
      "learning_rate": 0.0001394300486534617,
      "loss": 1.1554,
      "step": 5409
    },
    {
      "epoch": 0.38996612124270164,
      "grad_norm": 0.13695473130346403,
      "learning_rate": 0.0001394085919774237,
      "loss": 1.1108,
      "step": 5410
    },
    {
      "epoch": 0.39003820370503856,
      "grad_norm": 0.1183253218181648,
      "learning_rate": 0.0001393871331532649,
      "loss": 1.0816,
      "step": 5411
    },
    {
      "epoch": 0.3901102861673755,
      "grad_norm": 0.11070016316982277,
      "learning_rate": 0.00013936567218215498,
      "loss": 1.039,
      "step": 5412
    },
    {
      "epoch": 0.3901823686297124,
      "grad_norm": 0.10213199358947962,
      "learning_rate": 0.00013934420906526373,
      "loss": 1.0063,
      "step": 5413
    },
    {
      "epoch": 0.3902544510920493,
      "grad_norm": 0.1318402757489526,
      "learning_rate": 0.00013932274380376111,
      "loss": 1.0809,
      "step": 5414
    },
    {
      "epoch": 0.39032653355438623,
      "grad_norm": 0.12171087511256663,
      "learning_rate": 0.00013930127639881715,
      "loss": 1.0736,
      "step": 5415
    },
    {
      "epoch": 0.39039861601672315,
      "grad_norm": 0.13124258500163333,
      "learning_rate": 0.00013927980685160204,
      "loss": 1.099,
      "step": 5416
    },
    {
      "epoch": 0.39047069847906,
      "grad_norm": 0.12687202803102876,
      "learning_rate": 0.00013925833516328606,
      "loss": 1.0711,
      "step": 5417
    },
    {
      "epoch": 0.39054278094139694,
      "grad_norm": 0.13404306430394736,
      "learning_rate": 0.00013923686133503958,
      "loss": 1.0943,
      "step": 5418
    },
    {
      "epoch": 0.39061486340373386,
      "grad_norm": 0.13672517077760793,
      "learning_rate": 0.00013921538536803313,
      "loss": 1.0945,
      "step": 5419
    },
    {
      "epoch": 0.3906869458660708,
      "grad_norm": 0.14224580264178788,
      "learning_rate": 0.00013919390726343739,
      "loss": 1.0531,
      "step": 5420
    },
    {
      "epoch": 0.3907590283284077,
      "grad_norm": 0.121924462287195,
      "learning_rate": 0.00013917242702242303,
      "loss": 1.0772,
      "step": 5421
    },
    {
      "epoch": 0.3908311107907446,
      "grad_norm": 0.1722411662596467,
      "learning_rate": 0.00013915094464616096,
      "loss": 1.0739,
      "step": 5422
    },
    {
      "epoch": 0.39090319325308154,
      "grad_norm": 0.12643137145343708,
      "learning_rate": 0.00013912946013582212,
      "loss": 1.0776,
      "step": 5423
    },
    {
      "epoch": 0.39097527571541846,
      "grad_norm": 0.4247530125635977,
      "learning_rate": 0.0001391079734925777,
      "loss": 1.0541,
      "step": 5424
    },
    {
      "epoch": 0.3910473581777554,
      "grad_norm": 0.1671736667725767,
      "learning_rate": 0.00013908648471759878,
      "loss": 1.0571,
      "step": 5425
    },
    {
      "epoch": 0.39111944064009224,
      "grad_norm": 0.11192271163166331,
      "learning_rate": 0.00013906499381205684,
      "loss": 1.0669,
      "step": 5426
    },
    {
      "epoch": 0.39119152310242916,
      "grad_norm": 0.11733306884706639,
      "learning_rate": 0.0001390435007771232,
      "loss": 1.0207,
      "step": 5427
    },
    {
      "epoch": 0.3912636055647661,
      "grad_norm": 0.11439043253516414,
      "learning_rate": 0.0001390220056139695,
      "loss": 1.0554,
      "step": 5428
    },
    {
      "epoch": 0.391335688027103,
      "grad_norm": 0.10332160995279717,
      "learning_rate": 0.00013900050832376737,
      "loss": 1.0554,
      "step": 5429
    },
    {
      "epoch": 0.3914077704894399,
      "grad_norm": 0.1236679170230425,
      "learning_rate": 0.00013897900890768867,
      "loss": 1.0401,
      "step": 5430
    },
    {
      "epoch": 0.39147985295177684,
      "grad_norm": 0.10983075785756719,
      "learning_rate": 0.00013895750736690526,
      "loss": 1.0136,
      "step": 5431
    },
    {
      "epoch": 0.39155193541411376,
      "grad_norm": 0.42021631067947735,
      "learning_rate": 0.00013893600370258914,
      "loss": 1.0896,
      "step": 5432
    },
    {
      "epoch": 0.3916240178764507,
      "grad_norm": 0.17344049576092546,
      "learning_rate": 0.0001389144979159125,
      "loss": 1.0817,
      "step": 5433
    },
    {
      "epoch": 0.3916961003387876,
      "grad_norm": 0.3439781089259069,
      "learning_rate": 0.0001388929900080476,
      "loss": 1.0458,
      "step": 5434
    },
    {
      "epoch": 0.39176818280112446,
      "grad_norm": 0.11251064303379174,
      "learning_rate": 0.0001388714799801668,
      "loss": 1.0997,
      "step": 5435
    },
    {
      "epoch": 0.3918402652634614,
      "grad_norm": 0.10800293130769231,
      "learning_rate": 0.00013884996783344256,
      "loss": 0.9672,
      "step": 5436
    },
    {
      "epoch": 0.3919123477257983,
      "grad_norm": 0.12906880631578874,
      "learning_rate": 0.00013882845356904752,
      "loss": 1.0485,
      "step": 5437
    },
    {
      "epoch": 0.3919844301881352,
      "grad_norm": 0.15102962170865353,
      "learning_rate": 0.00013880693718815442,
      "loss": 1.0398,
      "step": 5438
    },
    {
      "epoch": 0.39205651265047214,
      "grad_norm": 0.20619277717893236,
      "learning_rate": 0.00013878541869193601,
      "loss": 0.9656,
      "step": 5439
    },
    {
      "epoch": 0.39212859511280906,
      "grad_norm": 0.10846018718650602,
      "learning_rate": 0.00013876389808156533,
      "loss": 1.063,
      "step": 5440
    },
    {
      "epoch": 0.392200677575146,
      "grad_norm": 0.11075702463608346,
      "learning_rate": 0.00013874237535821542,
      "loss": 1.1051,
      "step": 5441
    },
    {
      "epoch": 0.3922727600374829,
      "grad_norm": 0.1105559447415787,
      "learning_rate": 0.00013872085052305945,
      "loss": 1.0276,
      "step": 5442
    },
    {
      "epoch": 0.39234484249981977,
      "grad_norm": 0.10462997938893898,
      "learning_rate": 0.0001386993235772707,
      "loss": 1.0517,
      "step": 5443
    },
    {
      "epoch": 0.3924169249621567,
      "grad_norm": 0.10668569879591346,
      "learning_rate": 0.00013867779452202266,
      "loss": 1.0362,
      "step": 5444
    },
    {
      "epoch": 0.3924890074244936,
      "grad_norm": 0.11053285150228104,
      "learning_rate": 0.00013865626335848874,
      "loss": 1.145,
      "step": 5445
    },
    {
      "epoch": 0.3925610898868305,
      "grad_norm": 0.1123188911468739,
      "learning_rate": 0.00013863473008784264,
      "loss": 1.1032,
      "step": 5446
    },
    {
      "epoch": 0.39263317234916745,
      "grad_norm": 0.15905698677984198,
      "learning_rate": 0.0001386131947112581,
      "loss": 1.0801,
      "step": 5447
    },
    {
      "epoch": 0.39270525481150437,
      "grad_norm": 0.22182422605392904,
      "learning_rate": 0.0001385916572299091,
      "loss": 1.0817,
      "step": 5448
    },
    {
      "epoch": 0.3927773372738413,
      "grad_norm": 0.11201974038982483,
      "learning_rate": 0.00013857011764496942,
      "loss": 1.0421,
      "step": 5449
    },
    {
      "epoch": 0.3928494197361782,
      "grad_norm": 0.12036600698152965,
      "learning_rate": 0.00013854857595761334,
      "loss": 1.071,
      "step": 5450
    },
    {
      "epoch": 0.3929215021985151,
      "grad_norm": 0.10374117843826898,
      "learning_rate": 0.00013852703216901495,
      "loss": 1.0491,
      "step": 5451
    },
    {
      "epoch": 0.392993584660852,
      "grad_norm": 0.10837481832875066,
      "learning_rate": 0.00013850548628034867,
      "loss": 1.0619,
      "step": 5452
    },
    {
      "epoch": 0.3930656671231889,
      "grad_norm": 0.11914099746191444,
      "learning_rate": 0.0001384839382927889,
      "loss": 1.0227,
      "step": 5453
    },
    {
      "epoch": 0.39313774958552583,
      "grad_norm": 0.11258797491717733,
      "learning_rate": 0.0001384623882075102,
      "loss": 1.0976,
      "step": 5454
    },
    {
      "epoch": 0.39320983204786275,
      "grad_norm": 0.15091217908175897,
      "learning_rate": 0.00013844083602568725,
      "loss": 1.0805,
      "step": 5455
    },
    {
      "epoch": 0.39328191451019967,
      "grad_norm": 0.12423701596838492,
      "learning_rate": 0.00013841928174849485,
      "loss": 1.1182,
      "step": 5456
    },
    {
      "epoch": 0.3933539969725366,
      "grad_norm": 0.11754691956614592,
      "learning_rate": 0.00013839772537710786,
      "loss": 1.0607,
      "step": 5457
    },
    {
      "epoch": 0.3934260794348735,
      "grad_norm": 0.10590370100503672,
      "learning_rate": 0.00013837616691270136,
      "loss": 1.0757,
      "step": 5458
    },
    {
      "epoch": 0.39349816189721043,
      "grad_norm": 0.12281597994024224,
      "learning_rate": 0.00013835460635645039,
      "loss": 1.0611,
      "step": 5459
    },
    {
      "epoch": 0.39357024435954735,
      "grad_norm": 0.1100116090570164,
      "learning_rate": 0.00013833304370953028,
      "loss": 1.1126,
      "step": 5460
    },
    {
      "epoch": 0.3936423268218842,
      "grad_norm": 0.11675624330733823,
      "learning_rate": 0.00013831147897311633,
      "loss": 1.055,
      "step": 5461
    },
    {
      "epoch": 0.39371440928422113,
      "grad_norm": 0.14366062401098403,
      "learning_rate": 0.00013828991214838405,
      "loss": 1.0719,
      "step": 5462
    },
    {
      "epoch": 0.39378649174655805,
      "grad_norm": 0.12433551664363839,
      "learning_rate": 0.000138268343236509,
      "loss": 1.0476,
      "step": 5463
    },
    {
      "epoch": 0.393858574208895,
      "grad_norm": 0.11375953997537934,
      "learning_rate": 0.00013824677223866686,
      "loss": 1.0713,
      "step": 5464
    },
    {
      "epoch": 0.3939306566712319,
      "grad_norm": 0.11033592726131494,
      "learning_rate": 0.0001382251991560335,
      "loss": 1.062,
      "step": 5465
    },
    {
      "epoch": 0.3940027391335688,
      "grad_norm": 0.1088203389866428,
      "learning_rate": 0.0001382036239897848,
      "loss": 1.0266,
      "step": 5466
    },
    {
      "epoch": 0.39407482159590573,
      "grad_norm": 0.13972989092734148,
      "learning_rate": 0.00013818204674109682,
      "loss": 1.0903,
      "step": 5467
    },
    {
      "epoch": 0.39414690405824265,
      "grad_norm": 0.1319885351196477,
      "learning_rate": 0.0001381604674111457,
      "loss": 1.1056,
      "step": 5468
    },
    {
      "epoch": 0.3942189865205795,
      "grad_norm": 0.12209367740597085,
      "learning_rate": 0.0001381388860011077,
      "loss": 1.031,
      "step": 5469
    },
    {
      "epoch": 0.39429106898291644,
      "grad_norm": 0.1317613033075107,
      "learning_rate": 0.00013811730251215925,
      "loss": 1.1362,
      "step": 5470
    },
    {
      "epoch": 0.39436315144525336,
      "grad_norm": 0.11602663646988101,
      "learning_rate": 0.00013809571694547678,
      "loss": 0.9882,
      "step": 5471
    },
    {
      "epoch": 0.3944352339075903,
      "grad_norm": 0.1192049760329672,
      "learning_rate": 0.00013807412930223693,
      "loss": 1.1001,
      "step": 5472
    },
    {
      "epoch": 0.3945073163699272,
      "grad_norm": 0.1141881537611738,
      "learning_rate": 0.00013805253958361642,
      "loss": 1.0804,
      "step": 5473
    },
    {
      "epoch": 0.3945793988322641,
      "grad_norm": 0.11761502789355496,
      "learning_rate": 0.00013803094779079204,
      "loss": 1.0534,
      "step": 5474
    },
    {
      "epoch": 0.39465148129460104,
      "grad_norm": 0.10887703948538456,
      "learning_rate": 0.00013800935392494084,
      "loss": 1.0093,
      "step": 5475
    },
    {
      "epoch": 0.39472356375693796,
      "grad_norm": 0.10964418025373228,
      "learning_rate": 0.00013798775798723974,
      "loss": 1.0685,
      "step": 5476
    },
    {
      "epoch": 0.3947956462192749,
      "grad_norm": 0.15890089710014393,
      "learning_rate": 0.000137966159978866,
      "loss": 1.1134,
      "step": 5477
    },
    {
      "epoch": 0.39486772868161174,
      "grad_norm": 0.10105802137837737,
      "learning_rate": 0.00013794455990099688,
      "loss": 1.0806,
      "step": 5478
    },
    {
      "epoch": 0.39493981114394866,
      "grad_norm": 0.12148715264240802,
      "learning_rate": 0.00013792295775480983,
      "loss": 1.0479,
      "step": 5479
    },
    {
      "epoch": 0.3950118936062856,
      "grad_norm": 0.1134672718332191,
      "learning_rate": 0.00013790135354148226,
      "loss": 1.0405,
      "step": 5480
    },
    {
      "epoch": 0.3950839760686225,
      "grad_norm": 0.15345624260490856,
      "learning_rate": 0.0001378797472621919,
      "loss": 1.1275,
      "step": 5481
    },
    {
      "epoch": 0.3951560585309594,
      "grad_norm": 0.1139757976427702,
      "learning_rate": 0.00013785813891811636,
      "loss": 1.0682,
      "step": 5482
    },
    {
      "epoch": 0.39522814099329634,
      "grad_norm": 0.12078259555062144,
      "learning_rate": 0.00013783652851043362,
      "loss": 1.0445,
      "step": 5483
    },
    {
      "epoch": 0.39530022345563326,
      "grad_norm": 0.11396955599653164,
      "learning_rate": 0.00013781491604032152,
      "loss": 1.006,
      "step": 5484
    },
    {
      "epoch": 0.3953723059179702,
      "grad_norm": 0.11298397740123936,
      "learning_rate": 0.00013779330150895825,
      "loss": 1.152,
      "step": 5485
    },
    {
      "epoch": 0.39544438838030704,
      "grad_norm": 0.12429316787342695,
      "learning_rate": 0.00013777168491752194,
      "loss": 1.0233,
      "step": 5486
    },
    {
      "epoch": 0.39551647084264396,
      "grad_norm": 0.12389375857185511,
      "learning_rate": 0.00013775006626719086,
      "loss": 1.0524,
      "step": 5487
    },
    {
      "epoch": 0.3955885533049809,
      "grad_norm": 0.13385113773352555,
      "learning_rate": 0.00013772844555914347,
      "loss": 1.1064,
      "step": 5488
    },
    {
      "epoch": 0.3956606357673178,
      "grad_norm": 0.11301594627342317,
      "learning_rate": 0.00013770682279455823,
      "loss": 1.0793,
      "step": 5489
    },
    {
      "epoch": 0.3957327182296547,
      "grad_norm": 0.25953869852704275,
      "learning_rate": 0.00013768519797461386,
      "loss": 1.146,
      "step": 5490
    },
    {
      "epoch": 0.39580480069199164,
      "grad_norm": 0.24613127663106024,
      "learning_rate": 0.00013766357110048903,
      "loss": 1.0488,
      "step": 5491
    },
    {
      "epoch": 0.39587688315432856,
      "grad_norm": 0.18757637573095318,
      "learning_rate": 0.00013764194217336264,
      "loss": 1.1194,
      "step": 5492
    },
    {
      "epoch": 0.3959489656166655,
      "grad_norm": 0.160911253142499,
      "learning_rate": 0.00013762031119441366,
      "loss": 1.0023,
      "step": 5493
    },
    {
      "epoch": 0.3960210480790024,
      "grad_norm": 0.11963943281305366,
      "learning_rate": 0.00013759867816482113,
      "loss": 1.1054,
      "step": 5494
    },
    {
      "epoch": 0.39609313054133927,
      "grad_norm": 0.12407201887392916,
      "learning_rate": 0.00013757704308576426,
      "loss": 1.025,
      "step": 5495
    },
    {
      "epoch": 0.3961652130036762,
      "grad_norm": 0.15278979329102985,
      "learning_rate": 0.0001375554059584224,
      "loss": 1.0854,
      "step": 5496
    },
    {
      "epoch": 0.3962372954660131,
      "grad_norm": 0.12268210685977943,
      "learning_rate": 0.00013753376678397495,
      "loss": 1.1194,
      "step": 5497
    },
    {
      "epoch": 0.39630937792835,
      "grad_norm": 0.18304319902369295,
      "learning_rate": 0.00013751212556360138,
      "loss": 0.9981,
      "step": 5498
    },
    {
      "epoch": 0.39638146039068695,
      "grad_norm": 0.12120717247901674,
      "learning_rate": 0.0001374904822984814,
      "loss": 1.0047,
      "step": 5499
    },
    {
      "epoch": 0.39645354285302387,
      "grad_norm": 0.2985152847281636,
      "learning_rate": 0.00013746883698979474,
      "loss": 1.0814,
      "step": 5500
    },
    {
      "epoch": 0.3965256253153608,
      "grad_norm": 0.1329186212869446,
      "learning_rate": 0.00013744718963872124,
      "loss": 1.0736,
      "step": 5501
    },
    {
      "epoch": 0.3965977077776977,
      "grad_norm": 0.11561439478510847,
      "learning_rate": 0.0001374255402464409,
      "loss": 1.0566,
      "step": 5502
    },
    {
      "epoch": 0.3966697902400346,
      "grad_norm": 0.11562185368555625,
      "learning_rate": 0.0001374038888141338,
      "loss": 1.1522,
      "step": 5503
    },
    {
      "epoch": 0.3967418727023715,
      "grad_norm": 0.11015479206476357,
      "learning_rate": 0.00013738223534298012,
      "loss": 1.1052,
      "step": 5504
    },
    {
      "epoch": 0.3968139551647084,
      "grad_norm": 0.11725387742495572,
      "learning_rate": 0.00013736057983416023,
      "loss": 0.941,
      "step": 5505
    },
    {
      "epoch": 0.39688603762704533,
      "grad_norm": 0.12659188025916024,
      "learning_rate": 0.00013733892228885442,
      "loss": 1.1032,
      "step": 5506
    },
    {
      "epoch": 0.39695812008938225,
      "grad_norm": 0.11932465038788835,
      "learning_rate": 0.0001373172627082434,
      "loss": 1.0507,
      "step": 5507
    },
    {
      "epoch": 0.39703020255171917,
      "grad_norm": 0.10938140481294321,
      "learning_rate": 0.00013729560109350763,
      "loss": 1.0786,
      "step": 5508
    },
    {
      "epoch": 0.3971022850140561,
      "grad_norm": 0.27565354017048466,
      "learning_rate": 0.00013727393744582798,
      "loss": 1.0521,
      "step": 5509
    },
    {
      "epoch": 0.397174367476393,
      "grad_norm": 0.12330129682994552,
      "learning_rate": 0.00013725227176638528,
      "loss": 0.988,
      "step": 5510
    },
    {
      "epoch": 0.39724644993872993,
      "grad_norm": 0.11370992750724522,
      "learning_rate": 0.0001372306040563605,
      "loss": 1.1208,
      "step": 5511
    },
    {
      "epoch": 0.3973185324010668,
      "grad_norm": 0.11466072275426219,
      "learning_rate": 0.0001372089343169347,
      "loss": 1.0878,
      "step": 5512
    },
    {
      "epoch": 0.3973906148634037,
      "grad_norm": 0.8683304365781495,
      "learning_rate": 0.00013718726254928913,
      "loss": 1.0205,
      "step": 5513
    },
    {
      "epoch": 0.39746269732574063,
      "grad_norm": 0.1193399611800714,
      "learning_rate": 0.00013716558875460507,
      "loss": 1.0876,
      "step": 5514
    },
    {
      "epoch": 0.39753477978807755,
      "grad_norm": 0.12915054876544838,
      "learning_rate": 0.00013714391293406387,
      "loss": 1.101,
      "step": 5515
    },
    {
      "epoch": 0.3976068622504145,
      "grad_norm": 0.1096161576515537,
      "learning_rate": 0.00013712223508884717,
      "loss": 1.075,
      "step": 5516
    },
    {
      "epoch": 0.3976789447127514,
      "grad_norm": 0.1110287014330825,
      "learning_rate": 0.00013710055522013655,
      "loss": 1.0461,
      "step": 5517
    },
    {
      "epoch": 0.3977510271750883,
      "grad_norm": 0.11760027085412068,
      "learning_rate": 0.00013707887332911373,
      "loss": 1.0045,
      "step": 5518
    },
    {
      "epoch": 0.39782310963742523,
      "grad_norm": 0.1205850065579941,
      "learning_rate": 0.00013705718941696063,
      "loss": 1.0429,
      "step": 5519
    },
    {
      "epoch": 0.39789519209976215,
      "grad_norm": 0.11418931541313311,
      "learning_rate": 0.00013703550348485917,
      "loss": 1.0083,
      "step": 5520
    },
    {
      "epoch": 0.397967274562099,
      "grad_norm": 0.12160328231354743,
      "learning_rate": 0.00013701381553399145,
      "loss": 1.0623,
      "step": 5521
    },
    {
      "epoch": 0.39803935702443594,
      "grad_norm": 0.12401601671766735,
      "learning_rate": 0.0001369921255655396,
      "loss": 1.076,
      "step": 5522
    },
    {
      "epoch": 0.39811143948677286,
      "grad_norm": 0.16566131140229476,
      "learning_rate": 0.00013697043358068607,
      "loss": 1.0351,
      "step": 5523
    },
    {
      "epoch": 0.3981835219491098,
      "grad_norm": 0.11110606459203616,
      "learning_rate": 0.00013694873958061305,
      "loss": 1.1252,
      "step": 5524
    },
    {
      "epoch": 0.3982556044114467,
      "grad_norm": 0.2517617223229031,
      "learning_rate": 0.00013692704356650325,
      "loss": 1.0616,
      "step": 5525
    },
    {
      "epoch": 0.3983276868737836,
      "grad_norm": 0.10230850479148014,
      "learning_rate": 0.00013690534553953917,
      "loss": 0.9919,
      "step": 5526
    },
    {
      "epoch": 0.39839976933612054,
      "grad_norm": 0.1114366460977341,
      "learning_rate": 0.00013688364550090362,
      "loss": 1.0129,
      "step": 5527
    },
    {
      "epoch": 0.39847185179845745,
      "grad_norm": 0.16680958180376115,
      "learning_rate": 0.00013686194345177938,
      "loss": 1.0849,
      "step": 5528
    },
    {
      "epoch": 0.3985439342607944,
      "grad_norm": 0.13154420641888742,
      "learning_rate": 0.0001368402393933495,
      "loss": 1.1274,
      "step": 5529
    },
    {
      "epoch": 0.39861601672313124,
      "grad_norm": 0.14259030619934998,
      "learning_rate": 0.00013681853332679698,
      "loss": 1.0185,
      "step": 5530
    },
    {
      "epoch": 0.39868809918546816,
      "grad_norm": 0.13977589960985773,
      "learning_rate": 0.00013679682525330497,
      "loss": 1.0696,
      "step": 5531
    },
    {
      "epoch": 0.3987601816478051,
      "grad_norm": 0.2763333166077715,
      "learning_rate": 0.00013677511517405682,
      "loss": 0.9632,
      "step": 5532
    },
    {
      "epoch": 0.398832264110142,
      "grad_norm": 0.12016348377290406,
      "learning_rate": 0.00013675340309023587,
      "loss": 1.0241,
      "step": 5533
    },
    {
      "epoch": 0.3989043465724789,
      "grad_norm": 0.13149076913400523,
      "learning_rate": 0.00013673168900302566,
      "loss": 1.1244,
      "step": 5534
    },
    {
      "epoch": 0.39897642903481584,
      "grad_norm": 0.1149395575305486,
      "learning_rate": 0.00013670997291360978,
      "loss": 1.0897,
      "step": 5535
    },
    {
      "epoch": 0.39904851149715276,
      "grad_norm": 0.12382398188995415,
      "learning_rate": 0.00013668825482317197,
      "loss": 1.0526,
      "step": 5536
    },
    {
      "epoch": 0.3991205939594897,
      "grad_norm": 0.10526168619762359,
      "learning_rate": 0.00013666653473289606,
      "loss": 1.0123,
      "step": 5537
    },
    {
      "epoch": 0.39919267642182654,
      "grad_norm": 0.10736450523927943,
      "learning_rate": 0.00013664481264396596,
      "loss": 1.0148,
      "step": 5538
    },
    {
      "epoch": 0.39926475888416346,
      "grad_norm": 0.1399715624259303,
      "learning_rate": 0.00013662308855756574,
      "loss": 1.0455,
      "step": 5539
    },
    {
      "epoch": 0.3993368413465004,
      "grad_norm": 0.1206753563384885,
      "learning_rate": 0.00013660136247487957,
      "loss": 1.08,
      "step": 5540
    },
    {
      "epoch": 0.3994089238088373,
      "grad_norm": 0.1350630144968724,
      "learning_rate": 0.00013657963439709167,
      "loss": 1.1088,
      "step": 5541
    },
    {
      "epoch": 0.3994810062711742,
      "grad_norm": 0.1263111253322372,
      "learning_rate": 0.00013655790432538645,
      "loss": 0.9434,
      "step": 5542
    },
    {
      "epoch": 0.39955308873351114,
      "grad_norm": 0.10754221039896643,
      "learning_rate": 0.0001365361722609484,
      "loss": 1.006,
      "step": 5543
    },
    {
      "epoch": 0.39962517119584806,
      "grad_norm": 0.11220189350794398,
      "learning_rate": 0.0001365144382049621,
      "loss": 1.0718,
      "step": 5544
    },
    {
      "epoch": 0.399697253658185,
      "grad_norm": 0.1383059605030004,
      "learning_rate": 0.00013649270215861225,
      "loss": 1.0202,
      "step": 5545
    },
    {
      "epoch": 0.3997693361205219,
      "grad_norm": 0.10888579889687644,
      "learning_rate": 0.00013647096412308367,
      "loss": 1.0803,
      "step": 5546
    },
    {
      "epoch": 0.39984141858285877,
      "grad_norm": 0.18733665001069247,
      "learning_rate": 0.00013644922409956125,
      "loss": 1.1498,
      "step": 5547
    },
    {
      "epoch": 0.3999135010451957,
      "grad_norm": 0.19564554364769107,
      "learning_rate": 0.00013642748208923005,
      "loss": 1.0305,
      "step": 5548
    },
    {
      "epoch": 0.3999855835075326,
      "grad_norm": 0.4706840108301373,
      "learning_rate": 0.00013640573809327514,
      "loss": 1.0986,
      "step": 5549
    },
    {
      "epoch": 0.4000576659698695,
      "grad_norm": 0.1522148059976862,
      "learning_rate": 0.00013638399211288188,
      "loss": 1.0632,
      "step": 5550
    },
    {
      "epoch": 0.40012974843220644,
      "grad_norm": 0.13481955586251856,
      "learning_rate": 0.0001363622441492355,
      "loss": 1.07,
      "step": 5551
    },
    {
      "epoch": 0.40020183089454336,
      "grad_norm": 0.11191838523591234,
      "learning_rate": 0.00013634049420352152,
      "loss": 0.9892,
      "step": 5552
    },
    {
      "epoch": 0.4002739133568803,
      "grad_norm": 0.11837194641511602,
      "learning_rate": 0.0001363187422769255,
      "loss": 1.0461,
      "step": 5553
    },
    {
      "epoch": 0.4003459958192172,
      "grad_norm": 0.216511569093322,
      "learning_rate": 0.0001362969883706331,
      "loss": 0.9604,
      "step": 5554
    },
    {
      "epoch": 0.4004180782815541,
      "grad_norm": 0.12337871433418196,
      "learning_rate": 0.0001362752324858301,
      "loss": 1.0477,
      "step": 5555
    },
    {
      "epoch": 0.400490160743891,
      "grad_norm": 0.13378035111648734,
      "learning_rate": 0.00013625347462370243,
      "loss": 1.0983,
      "step": 5556
    },
    {
      "epoch": 0.4005622432062279,
      "grad_norm": 0.20431484025688726,
      "learning_rate": 0.00013623171478543603,
      "loss": 1.0578,
      "step": 5557
    },
    {
      "epoch": 0.40063432566856483,
      "grad_norm": 0.10581195056006648,
      "learning_rate": 0.00013620995297221707,
      "loss": 1.0365,
      "step": 5558
    },
    {
      "epoch": 0.40070640813090175,
      "grad_norm": 0.11361295783735482,
      "learning_rate": 0.0001361881891852317,
      "loss": 1.0551,
      "step": 5559
    },
    {
      "epoch": 0.40077849059323867,
      "grad_norm": 0.12644101967237845,
      "learning_rate": 0.0001361664234256663,
      "loss": 1.1093,
      "step": 5560
    },
    {
      "epoch": 0.4008505730555756,
      "grad_norm": 0.1161659236651935,
      "learning_rate": 0.00013614465569470723,
      "loss": 1.0546,
      "step": 5561
    },
    {
      "epoch": 0.4009226555179125,
      "grad_norm": 0.13093146417726464,
      "learning_rate": 0.00013612288599354112,
      "loss": 1.0291,
      "step": 5562
    },
    {
      "epoch": 0.4009947379802494,
      "grad_norm": 0.1336646763713269,
      "learning_rate": 0.00013610111432335454,
      "loss": 1.014,
      "step": 5563
    },
    {
      "epoch": 0.4010668204425863,
      "grad_norm": 0.12664587589558204,
      "learning_rate": 0.00013607934068533424,
      "loss": 1.15,
      "step": 5564
    },
    {
      "epoch": 0.4011389029049232,
      "grad_norm": 0.12888870765447477,
      "learning_rate": 0.0001360575650806671,
      "loss": 1.0437,
      "step": 5565
    },
    {
      "epoch": 0.40121098536726013,
      "grad_norm": 0.131820555708967,
      "learning_rate": 0.0001360357875105401,
      "loss": 1.0974,
      "step": 5566
    },
    {
      "epoch": 0.40128306782959705,
      "grad_norm": 0.12171065242778203,
      "learning_rate": 0.00013601400797614029,
      "loss": 1.1036,
      "step": 5567
    },
    {
      "epoch": 0.40135515029193397,
      "grad_norm": 0.10567612473879713,
      "learning_rate": 0.00013599222647865485,
      "loss": 1.045,
      "step": 5568
    },
    {
      "epoch": 0.4014272327542709,
      "grad_norm": 0.1352915332312915,
      "learning_rate": 0.00013597044301927105,
      "loss": 1.0354,
      "step": 5569
    },
    {
      "epoch": 0.4014993152166078,
      "grad_norm": 0.12764254150850604,
      "learning_rate": 0.00013594865759917633,
      "loss": 1.0332,
      "step": 5570
    },
    {
      "epoch": 0.40157139767894473,
      "grad_norm": 0.14256661319464353,
      "learning_rate": 0.00013592687021955814,
      "loss": 1.0784,
      "step": 5571
    },
    {
      "epoch": 0.40164348014128165,
      "grad_norm": 0.1302207335148006,
      "learning_rate": 0.00013590508088160414,
      "loss": 1.1745,
      "step": 5572
    },
    {
      "epoch": 0.4017155626036185,
      "grad_norm": 0.13474820148358765,
      "learning_rate": 0.000135883289586502,
      "loss": 1.0509,
      "step": 5573
    },
    {
      "epoch": 0.40178764506595543,
      "grad_norm": 0.11744144010292841,
      "learning_rate": 0.00013586149633543958,
      "loss": 1.0928,
      "step": 5574
    },
    {
      "epoch": 0.40185972752829235,
      "grad_norm": 0.12847141058108366,
      "learning_rate": 0.00013583970112960474,
      "loss": 0.9913,
      "step": 5575
    },
    {
      "epoch": 0.4019318099906293,
      "grad_norm": 0.10549555647065303,
      "learning_rate": 0.00013581790397018562,
      "loss": 1.102,
      "step": 5576
    },
    {
      "epoch": 0.4020038924529662,
      "grad_norm": 0.11634910287763042,
      "learning_rate": 0.00013579610485837024,
      "loss": 1.0745,
      "step": 5577
    },
    {
      "epoch": 0.4020759749153031,
      "grad_norm": 0.14785898031633127,
      "learning_rate": 0.00013577430379534694,
      "loss": 1.1052,
      "step": 5578
    },
    {
      "epoch": 0.40214805737764003,
      "grad_norm": 0.1305300426809869,
      "learning_rate": 0.000135752500782304,
      "loss": 1.0142,
      "step": 5579
    },
    {
      "epoch": 0.40222013983997695,
      "grad_norm": 0.11904494181415957,
      "learning_rate": 0.00013573069582042998,
      "loss": 1.1211,
      "step": 5580
    },
    {
      "epoch": 0.4022922223023139,
      "grad_norm": 0.13239264484455784,
      "learning_rate": 0.0001357088889109133,
      "loss": 1.078,
      "step": 5581
    },
    {
      "epoch": 0.40236430476465074,
      "grad_norm": 0.1673008459462315,
      "learning_rate": 0.00013568708005494277,
      "loss": 1.06,
      "step": 5582
    },
    {
      "epoch": 0.40243638722698766,
      "grad_norm": 0.1270430624049211,
      "learning_rate": 0.0001356652692537071,
      "loss": 1.0383,
      "step": 5583
    },
    {
      "epoch": 0.4025084696893246,
      "grad_norm": 0.12393314062525584,
      "learning_rate": 0.00013564345650839517,
      "loss": 1.0968,
      "step": 5584
    },
    {
      "epoch": 0.4025805521516615,
      "grad_norm": 0.15676962613279663,
      "learning_rate": 0.000135621641820196,
      "loss": 1.0121,
      "step": 5585
    },
    {
      "epoch": 0.4026526346139984,
      "grad_norm": 0.14063839145208717,
      "learning_rate": 0.0001355998251902987,
      "loss": 1.0504,
      "step": 5586
    },
    {
      "epoch": 0.40272471707633534,
      "grad_norm": 0.11555401604942435,
      "learning_rate": 0.0001355780066198924,
      "loss": 1.0911,
      "step": 5587
    },
    {
      "epoch": 0.40279679953867226,
      "grad_norm": 0.12496560445281098,
      "learning_rate": 0.00013555618611016645,
      "loss": 1.0286,
      "step": 5588
    },
    {
      "epoch": 0.4028688820010092,
      "grad_norm": 0.11288595584445574,
      "learning_rate": 0.00013553436366231026,
      "loss": 0.9917,
      "step": 5589
    },
    {
      "epoch": 0.40294096446334604,
      "grad_norm": 0.10767035031019823,
      "learning_rate": 0.00013551253927751334,
      "loss": 1.1323,
      "step": 5590
    },
    {
      "epoch": 0.40301304692568296,
      "grad_norm": 0.20051874454783472,
      "learning_rate": 0.00013549071295696535,
      "loss": 1.0595,
      "step": 5591
    },
    {
      "epoch": 0.4030851293880199,
      "grad_norm": 0.12924943384063498,
      "learning_rate": 0.00013546888470185602,
      "loss": 1.0412,
      "step": 5592
    },
    {
      "epoch": 0.4031572118503568,
      "grad_norm": 0.23015668090766733,
      "learning_rate": 0.00013544705451337512,
      "loss": 1.081,
      "step": 5593
    },
    {
      "epoch": 0.4032292943126937,
      "grad_norm": 0.12473586089871692,
      "learning_rate": 0.00013542522239271264,
      "loss": 1.0682,
      "step": 5594
    },
    {
      "epoch": 0.40330137677503064,
      "grad_norm": 0.18398807301698203,
      "learning_rate": 0.00013540338834105861,
      "loss": 0.9658,
      "step": 5595
    },
    {
      "epoch": 0.40337345923736756,
      "grad_norm": 0.11497578958383758,
      "learning_rate": 0.00013538155235960322,
      "loss": 1.0756,
      "step": 5596
    },
    {
      "epoch": 0.4034455416997045,
      "grad_norm": 0.11580590776870826,
      "learning_rate": 0.00013535971444953666,
      "loss": 1.0883,
      "step": 5597
    },
    {
      "epoch": 0.4035176241620414,
      "grad_norm": 0.3550625158230747,
      "learning_rate": 0.00013533787461204934,
      "loss": 1.0721,
      "step": 5598
    },
    {
      "epoch": 0.40358970662437826,
      "grad_norm": 0.5983836617872683,
      "learning_rate": 0.0001353160328483317,
      "loss": 1.0562,
      "step": 5599
    },
    {
      "epoch": 0.4036617890867152,
      "grad_norm": 0.11424745677082507,
      "learning_rate": 0.00013529418915957434,
      "loss": 1.0019,
      "step": 5600
    },
    {
      "epoch": 0.4037338715490521,
      "grad_norm": 0.27954639693907757,
      "learning_rate": 0.00013527234354696787,
      "loss": 1.0331,
      "step": 5601
    },
    {
      "epoch": 0.403805954011389,
      "grad_norm": 0.14816072089070187,
      "learning_rate": 0.00013525049601170318,
      "loss": 1.1009,
      "step": 5602
    },
    {
      "epoch": 0.40387803647372594,
      "grad_norm": 0.10836318324162425,
      "learning_rate": 0.00013522864655497106,
      "loss": 0.9751,
      "step": 5603
    },
    {
      "epoch": 0.40395011893606286,
      "grad_norm": 0.10914638107201736,
      "learning_rate": 0.00013520679517796259,
      "loss": 1.0654,
      "step": 5604
    },
    {
      "epoch": 0.4040222013983998,
      "grad_norm": 0.14530195437432197,
      "learning_rate": 0.00013518494188186878,
      "loss": 0.9944,
      "step": 5605
    },
    {
      "epoch": 0.4040942838607367,
      "grad_norm": 0.12567326091949427,
      "learning_rate": 0.00013516308666788087,
      "loss": 1.0354,
      "step": 5606
    },
    {
      "epoch": 0.4041663663230736,
      "grad_norm": 0.1062566714011408,
      "learning_rate": 0.00013514122953719014,
      "loss": 1.0153,
      "step": 5607
    },
    {
      "epoch": 0.4042384487854105,
      "grad_norm": 0.10978968707354171,
      "learning_rate": 0.00013511937049098805,
      "loss": 1.0224,
      "step": 5608
    },
    {
      "epoch": 0.4043105312477474,
      "grad_norm": 0.1418691616401946,
      "learning_rate": 0.00013509750953046606,
      "loss": 1.055,
      "step": 5609
    },
    {
      "epoch": 0.4043826137100843,
      "grad_norm": 0.461838735706658,
      "learning_rate": 0.0001350756466568158,
      "loss": 1.1133,
      "step": 5610
    },
    {
      "epoch": 0.40445469617242125,
      "grad_norm": 0.1139250866536068,
      "learning_rate": 0.00013505378187122903,
      "loss": 0.9764,
      "step": 5611
    },
    {
      "epoch": 0.40452677863475817,
      "grad_norm": 0.12257869725680783,
      "learning_rate": 0.0001350319151748975,
      "loss": 1.0719,
      "step": 5612
    },
    {
      "epoch": 0.4045988610970951,
      "grad_norm": 0.1251169970046012,
      "learning_rate": 0.00013501004656901324,
      "loss": 1.1185,
      "step": 5613
    },
    {
      "epoch": 0.404670943559432,
      "grad_norm": 0.3771176467441942,
      "learning_rate": 0.0001349881760547682,
      "loss": 1.112,
      "step": 5614
    },
    {
      "epoch": 0.4047430260217689,
      "grad_norm": 0.11651454894634761,
      "learning_rate": 0.00013496630363335455,
      "loss": 1.0431,
      "step": 5615
    },
    {
      "epoch": 0.4048151084841058,
      "grad_norm": 0.1331530286562252,
      "learning_rate": 0.00013494442930596457,
      "loss": 1.0801,
      "step": 5616
    },
    {
      "epoch": 0.4048871909464427,
      "grad_norm": 0.10260291403838466,
      "learning_rate": 0.00013492255307379056,
      "loss": 1.0773,
      "step": 5617
    },
    {
      "epoch": 0.40495927340877963,
      "grad_norm": 0.11799821802837626,
      "learning_rate": 0.00013490067493802496,
      "loss": 1.0905,
      "step": 5618
    },
    {
      "epoch": 0.40503135587111655,
      "grad_norm": 0.9532542476246861,
      "learning_rate": 0.00013487879489986035,
      "loss": 1.0151,
      "step": 5619
    },
    {
      "epoch": 0.40510343833345347,
      "grad_norm": 0.1268334350951427,
      "learning_rate": 0.0001348569129604894,
      "loss": 1.0503,
      "step": 5620
    },
    {
      "epoch": 0.4051755207957904,
      "grad_norm": 0.13970424178487817,
      "learning_rate": 0.00013483502912110484,
      "loss": 1.1776,
      "step": 5621
    },
    {
      "epoch": 0.4052476032581273,
      "grad_norm": 0.7864022416748205,
      "learning_rate": 0.00013481314338289956,
      "loss": 1.1068,
      "step": 5622
    },
    {
      "epoch": 0.40531968572046423,
      "grad_norm": 0.12618352171042452,
      "learning_rate": 0.00013479125574706653,
      "loss": 1.0672,
      "step": 5623
    },
    {
      "epoch": 0.40539176818280115,
      "grad_norm": 0.16472148187357855,
      "learning_rate": 0.00013476936621479877,
      "loss": 1.017,
      "step": 5624
    },
    {
      "epoch": 0.405463850645138,
      "grad_norm": 0.12160984557573194,
      "learning_rate": 0.00013474747478728954,
      "loss": 1.0339,
      "step": 5625
    },
    {
      "epoch": 0.40553593310747493,
      "grad_norm": 0.12688722896918483,
      "learning_rate": 0.00013472558146573204,
      "loss": 1.1099,
      "step": 5626
    },
    {
      "epoch": 0.40560801556981185,
      "grad_norm": 0.16090454366454723,
      "learning_rate": 0.0001347036862513197,
      "loss": 1.0727,
      "step": 5627
    },
    {
      "epoch": 0.4056800980321488,
      "grad_norm": 0.1127877100023557,
      "learning_rate": 0.000134681789145246,
      "loss": 1.0487,
      "step": 5628
    },
    {
      "epoch": 0.4057521804944857,
      "grad_norm": 0.12855610669222584,
      "learning_rate": 0.00013465989014870453,
      "loss": 1.1136,
      "step": 5629
    },
    {
      "epoch": 0.4058242629568226,
      "grad_norm": 0.1461746394971455,
      "learning_rate": 0.00013463798926288892,
      "loss": 1.0192,
      "step": 5630
    },
    {
      "epoch": 0.40589634541915953,
      "grad_norm": 0.12737212236793885,
      "learning_rate": 0.00013461608648899308,
      "loss": 1.0175,
      "step": 5631
    },
    {
      "epoch": 0.40596842788149645,
      "grad_norm": 0.2760241245421851,
      "learning_rate": 0.00013459418182821078,
      "loss": 1.0654,
      "step": 5632
    },
    {
      "epoch": 0.40604051034383337,
      "grad_norm": 0.11116883958767218,
      "learning_rate": 0.00013457227528173614,
      "loss": 1.0928,
      "step": 5633
    },
    {
      "epoch": 0.40611259280617024,
      "grad_norm": 0.13499940091448415,
      "learning_rate": 0.00013455036685076314,
      "loss": 1.0628,
      "step": 5634
    },
    {
      "epoch": 0.40618467526850716,
      "grad_norm": 0.12617925876763705,
      "learning_rate": 0.0001345284565364861,
      "loss": 1.029,
      "step": 5635
    },
    {
      "epoch": 0.4062567577308441,
      "grad_norm": 0.14220381171029003,
      "learning_rate": 0.00013450654434009923,
      "loss": 1.0533,
      "step": 5636
    },
    {
      "epoch": 0.406328840193181,
      "grad_norm": 0.10256116499079702,
      "learning_rate": 0.00013448463026279704,
      "loss": 1.0169,
      "step": 5637
    },
    {
      "epoch": 0.4064009226555179,
      "grad_norm": 0.133889379615737,
      "learning_rate": 0.000134462714305774,
      "loss": 1.1327,
      "step": 5638
    },
    {
      "epoch": 0.40647300511785484,
      "grad_norm": 0.2425506831459239,
      "learning_rate": 0.0001344407964702247,
      "loss": 1.0548,
      "step": 5639
    },
    {
      "epoch": 0.40654508758019176,
      "grad_norm": 0.12010257812357704,
      "learning_rate": 0.00013441887675734386,
      "loss": 1.0838,
      "step": 5640
    },
    {
      "epoch": 0.4066171700425287,
      "grad_norm": 0.12271531489452575,
      "learning_rate": 0.0001343969551683263,
      "loss": 1.0819,
      "step": 5641
    },
    {
      "epoch": 0.40668925250486554,
      "grad_norm": 0.11028169071977291,
      "learning_rate": 0.000134375031704367,
      "loss": 1.0172,
      "step": 5642
    },
    {
      "epoch": 0.40676133496720246,
      "grad_norm": 0.1880048763160138,
      "learning_rate": 0.00013435310636666093,
      "loss": 1.0729,
      "step": 5643
    },
    {
      "epoch": 0.4068334174295394,
      "grad_norm": 0.1244704610608357,
      "learning_rate": 0.00013433117915640324,
      "loss": 0.9787,
      "step": 5644
    },
    {
      "epoch": 0.4069054998918763,
      "grad_norm": 0.7630894706638466,
      "learning_rate": 0.00013430925007478913,
      "loss": 1.0805,
      "step": 5645
    },
    {
      "epoch": 0.4069775823542132,
      "grad_norm": 0.11513576401447752,
      "learning_rate": 0.00013428731912301398,
      "loss": 1.022,
      "step": 5646
    },
    {
      "epoch": 0.40704966481655014,
      "grad_norm": 0.17011670033004192,
      "learning_rate": 0.00013426538630227316,
      "loss": 1.0784,
      "step": 5647
    },
    {
      "epoch": 0.40712174727888706,
      "grad_norm": 0.15968975972001145,
      "learning_rate": 0.00013424345161376226,
      "loss": 1.1051,
      "step": 5648
    },
    {
      "epoch": 0.407193829741224,
      "grad_norm": 0.10458418972481251,
      "learning_rate": 0.00013422151505867693,
      "loss": 1.0441,
      "step": 5649
    },
    {
      "epoch": 0.4072659122035609,
      "grad_norm": 0.13043314104090076,
      "learning_rate": 0.0001341995766382128,
      "loss": 1.0453,
      "step": 5650
    },
    {
      "epoch": 0.40733799466589776,
      "grad_norm": 0.17700528706024576,
      "learning_rate": 0.00013417763635356584,
      "loss": 1.1001,
      "step": 5651
    },
    {
      "epoch": 0.4074100771282347,
      "grad_norm": 0.12183876149887969,
      "learning_rate": 0.0001341556942059319,
      "loss": 1.0424,
      "step": 5652
    },
    {
      "epoch": 0.4074821595905716,
      "grad_norm": 1.5030569073597448,
      "learning_rate": 0.0001341337501965071,
      "loss": 1.0902,
      "step": 5653
    },
    {
      "epoch": 0.4075542420529085,
      "grad_norm": 0.12569014960149447,
      "learning_rate": 0.0001341118043264875,
      "loss": 1.0496,
      "step": 5654
    },
    {
      "epoch": 0.40762632451524544,
      "grad_norm": 0.19500650118564916,
      "learning_rate": 0.00013408985659706945,
      "loss": 1.0531,
      "step": 5655
    },
    {
      "epoch": 0.40769840697758236,
      "grad_norm": 2.3415258203292124,
      "learning_rate": 0.00013406790700944918,
      "loss": 1.0256,
      "step": 5656
    },
    {
      "epoch": 0.4077704894399193,
      "grad_norm": 0.1267081843540592,
      "learning_rate": 0.00013404595556482324,
      "loss": 1.0612,
      "step": 5657
    },
    {
      "epoch": 0.4078425719022562,
      "grad_norm": 0.13736134871316139,
      "learning_rate": 0.00013402400226438812,
      "loss": 1.1298,
      "step": 5658
    },
    {
      "epoch": 0.4079146543645931,
      "grad_norm": 0.11884364165691062,
      "learning_rate": 0.0001340020471093405,
      "loss": 1.0496,
      "step": 5659
    },
    {
      "epoch": 0.40798673682693,
      "grad_norm": 0.15303049192491391,
      "learning_rate": 0.0001339800901008771,
      "loss": 1.02,
      "step": 5660
    },
    {
      "epoch": 0.4080588192892669,
      "grad_norm": 0.13467273907027227,
      "learning_rate": 0.00013395813124019484,
      "loss": 1.0899,
      "step": 5661
    },
    {
      "epoch": 0.4081309017516038,
      "grad_norm": 0.21436707610074593,
      "learning_rate": 0.00013393617052849058,
      "loss": 1.0625,
      "step": 5662
    },
    {
      "epoch": 0.40820298421394074,
      "grad_norm": 0.1348197649313236,
      "learning_rate": 0.00013391420796696145,
      "loss": 1.0967,
      "step": 5663
    },
    {
      "epoch": 0.40827506667627766,
      "grad_norm": 0.1308247057287976,
      "learning_rate": 0.00013389224355680454,
      "loss": 1.0453,
      "step": 5664
    },
    {
      "epoch": 0.4083471491386146,
      "grad_norm": 0.12711753783662325,
      "learning_rate": 0.0001338702772992172,
      "loss": 1.0849,
      "step": 5665
    },
    {
      "epoch": 0.4084192316009515,
      "grad_norm": 0.12274109053746933,
      "learning_rate": 0.0001338483091953967,
      "loss": 1.0854,
      "step": 5666
    },
    {
      "epoch": 0.4084913140632884,
      "grad_norm": 0.17695609152652306,
      "learning_rate": 0.00013382633924654056,
      "loss": 1.0652,
      "step": 5667
    },
    {
      "epoch": 0.4085633965256253,
      "grad_norm": 0.16846480444123274,
      "learning_rate": 0.00013380436745384628,
      "loss": 1.0725,
      "step": 5668
    },
    {
      "epoch": 0.4086354789879622,
      "grad_norm": 0.11309351184016438,
      "learning_rate": 0.0001337823938185116,
      "loss": 1.0339,
      "step": 5669
    },
    {
      "epoch": 0.40870756145029913,
      "grad_norm": 0.19322751844596675,
      "learning_rate": 0.00013376041834173419,
      "loss": 1.0526,
      "step": 5670
    },
    {
      "epoch": 0.40877964391263605,
      "grad_norm": 0.13319082647084923,
      "learning_rate": 0.00013373844102471198,
      "loss": 1.0792,
      "step": 5671
    },
    {
      "epoch": 0.40885172637497297,
      "grad_norm": 0.16783254770960343,
      "learning_rate": 0.00013371646186864286,
      "loss": 1.0757,
      "step": 5672
    },
    {
      "epoch": 0.4089238088373099,
      "grad_norm": 0.11607701510742129,
      "learning_rate": 0.00013369448087472498,
      "loss": 0.9699,
      "step": 5673
    },
    {
      "epoch": 0.4089958912996468,
      "grad_norm": 0.1542711604121803,
      "learning_rate": 0.0001336724980441564,
      "loss": 1.0475,
      "step": 5674
    },
    {
      "epoch": 0.4090679737619837,
      "grad_norm": 0.23774096659956523,
      "learning_rate": 0.0001336505133781355,
      "loss": 0.9645,
      "step": 5675
    },
    {
      "epoch": 0.40914005622432065,
      "grad_norm": 0.10392067322143422,
      "learning_rate": 0.0001336285268778605,
      "loss": 1.0508,
      "step": 5676
    },
    {
      "epoch": 0.4092121386866575,
      "grad_norm": 0.11771774423302715,
      "learning_rate": 0.00013360653854453,
      "loss": 1.0156,
      "step": 5677
    },
    {
      "epoch": 0.40928422114899443,
      "grad_norm": 0.12806533935890685,
      "learning_rate": 0.00013358454837934244,
      "loss": 1.0597,
      "step": 5678
    },
    {
      "epoch": 0.40935630361133135,
      "grad_norm": 0.2196325370362922,
      "learning_rate": 0.0001335625563834966,
      "loss": 0.8967,
      "step": 5679
    },
    {
      "epoch": 0.40942838607366827,
      "grad_norm": 0.12292689993144912,
      "learning_rate": 0.0001335405625581911,
      "loss": 1.1113,
      "step": 5680
    },
    {
      "epoch": 0.4095004685360052,
      "grad_norm": 0.12372307877464264,
      "learning_rate": 0.00013351856690462497,
      "loss": 1.1028,
      "step": 5681
    },
    {
      "epoch": 0.4095725509983421,
      "grad_norm": 0.14960414596674665,
      "learning_rate": 0.00013349656942399703,
      "loss": 1.0249,
      "step": 5682
    },
    {
      "epoch": 0.40964463346067903,
      "grad_norm": 0.10575782423206075,
      "learning_rate": 0.00013347457011750636,
      "loss": 0.9936,
      "step": 5683
    },
    {
      "epoch": 0.40971671592301595,
      "grad_norm": 0.11373511962615306,
      "learning_rate": 0.0001334525689863522,
      "loss": 1.0947,
      "step": 5684
    },
    {
      "epoch": 0.40978879838535287,
      "grad_norm": 0.12563457094933864,
      "learning_rate": 0.00013343056603173375,
      "loss": 1.0186,
      "step": 5685
    },
    {
      "epoch": 0.40986088084768973,
      "grad_norm": 0.13234263473347527,
      "learning_rate": 0.00013340856125485033,
      "loss": 0.999,
      "step": 5686
    },
    {
      "epoch": 0.40993296331002665,
      "grad_norm": 0.10570642588650245,
      "learning_rate": 0.00013338655465690148,
      "loss": 1.0652,
      "step": 5687
    },
    {
      "epoch": 0.4100050457723636,
      "grad_norm": 0.11151317468985045,
      "learning_rate": 0.00013336454623908667,
      "loss": 1.1254,
      "step": 5688
    },
    {
      "epoch": 0.4100771282347005,
      "grad_norm": 0.17413675692641306,
      "learning_rate": 0.00013334253600260563,
      "loss": 1.0598,
      "step": 5689
    },
    {
      "epoch": 0.4101492106970374,
      "grad_norm": 0.22666737521747507,
      "learning_rate": 0.0001333205239486581,
      "loss": 0.9755,
      "step": 5690
    },
    {
      "epoch": 0.41022129315937433,
      "grad_norm": 0.13706231193043147,
      "learning_rate": 0.00013329851007844393,
      "loss": 1.0754,
      "step": 5691
    },
    {
      "epoch": 0.41029337562171125,
      "grad_norm": 0.23674799605861976,
      "learning_rate": 0.00013327649439316304,
      "loss": 1.0955,
      "step": 5692
    },
    {
      "epoch": 0.4103654580840482,
      "grad_norm": 0.11551751266533032,
      "learning_rate": 0.00013325447689401548,
      "loss": 1.0499,
      "step": 5693
    },
    {
      "epoch": 0.41043754054638504,
      "grad_norm": 0.13512381036126211,
      "learning_rate": 0.00013323245758220146,
      "loss": 1.0783,
      "step": 5694
    },
    {
      "epoch": 0.41050962300872196,
      "grad_norm": 0.11372673772144551,
      "learning_rate": 0.0001332104364589212,
      "loss": 0.9612,
      "step": 5695
    },
    {
      "epoch": 0.4105817054710589,
      "grad_norm": 0.1193937827599972,
      "learning_rate": 0.00013318841352537503,
      "loss": 0.997,
      "step": 5696
    },
    {
      "epoch": 0.4106537879333958,
      "grad_norm": 0.17426096890919276,
      "learning_rate": 0.0001331663887827634,
      "loss": 1.0802,
      "step": 5697
    },
    {
      "epoch": 0.4107258703957327,
      "grad_norm": 0.11731730106653891,
      "learning_rate": 0.0001331443622322869,
      "loss": 1.0716,
      "step": 5698
    },
    {
      "epoch": 0.41079795285806964,
      "grad_norm": 0.12011446604743954,
      "learning_rate": 0.00013312233387514612,
      "loss": 1.0311,
      "step": 5699
    },
    {
      "epoch": 0.41087003532040656,
      "grad_norm": 0.11632845003863769,
      "learning_rate": 0.00013310030371254183,
      "loss": 1.0365,
      "step": 5700
    },
    {
      "epoch": 0.4109421177827435,
      "grad_norm": 0.11454370808752587,
      "learning_rate": 0.00013307827174567486,
      "loss": 1.1371,
      "step": 5701
    },
    {
      "epoch": 0.4110142002450804,
      "grad_norm": 0.1162588549935865,
      "learning_rate": 0.00013305623797574615,
      "loss": 0.9853,
      "step": 5702
    },
    {
      "epoch": 0.41108628270741726,
      "grad_norm": 0.2805801108587071,
      "learning_rate": 0.00013303420240395674,
      "loss": 1.0425,
      "step": 5703
    },
    {
      "epoch": 0.4111583651697542,
      "grad_norm": 0.19632769227567584,
      "learning_rate": 0.0001330121650315078,
      "loss": 1.063,
      "step": 5704
    },
    {
      "epoch": 0.4112304476320911,
      "grad_norm": 0.12431248761337102,
      "learning_rate": 0.00013299012585960048,
      "loss": 1.0785,
      "step": 5705
    },
    {
      "epoch": 0.411302530094428,
      "grad_norm": 0.7383152421709849,
      "learning_rate": 0.00013296808488943618,
      "loss": 1.0598,
      "step": 5706
    },
    {
      "epoch": 0.41137461255676494,
      "grad_norm": 0.12878792824939134,
      "learning_rate": 0.0001329460421222163,
      "loss": 1.0223,
      "step": 5707
    },
    {
      "epoch": 0.41144669501910186,
      "grad_norm": 0.11991947972128815,
      "learning_rate": 0.00013292399755914246,
      "loss": 1.0886,
      "step": 5708
    },
    {
      "epoch": 0.4115187774814388,
      "grad_norm": 0.10806560184263517,
      "learning_rate": 0.00013290195120141607,
      "loss": 1.0803,
      "step": 5709
    },
    {
      "epoch": 0.4115908599437757,
      "grad_norm": 0.14023633234140534,
      "learning_rate": 0.0001328799030502391,
      "loss": 1.0415,
      "step": 5710
    },
    {
      "epoch": 0.4116629424061126,
      "grad_norm": 0.12356981996642911,
      "learning_rate": 0.0001328578531068132,
      "loss": 1.0042,
      "step": 5711
    },
    {
      "epoch": 0.4117350248684495,
      "grad_norm": 0.10857538236024931,
      "learning_rate": 0.0001328358013723404,
      "loss": 1.067,
      "step": 5712
    },
    {
      "epoch": 0.4118071073307864,
      "grad_norm": 0.12668101882460558,
      "learning_rate": 0.00013281374784802262,
      "loss": 1.0587,
      "step": 5713
    },
    {
      "epoch": 0.4118791897931233,
      "grad_norm": 0.11546385495093801,
      "learning_rate": 0.00013279169253506206,
      "loss": 0.9616,
      "step": 5714
    },
    {
      "epoch": 0.41195127225546024,
      "grad_norm": 0.1906367677503871,
      "learning_rate": 0.0001327696354346609,
      "loss": 1.013,
      "step": 5715
    },
    {
      "epoch": 0.41202335471779716,
      "grad_norm": 0.13298367772391548,
      "learning_rate": 0.0001327475765480214,
      "loss": 1.0368,
      "step": 5716
    },
    {
      "epoch": 0.4120954371801341,
      "grad_norm": 0.38097311489689517,
      "learning_rate": 0.00013272551587634605,
      "loss": 1.0995,
      "step": 5717
    },
    {
      "epoch": 0.412167519642471,
      "grad_norm": 0.1301165875979599,
      "learning_rate": 0.0001327034534208373,
      "loss": 1.0437,
      "step": 5718
    },
    {
      "epoch": 0.4122396021048079,
      "grad_norm": 0.12393282581047882,
      "learning_rate": 0.00013268138918269776,
      "loss": 1.0217,
      "step": 5719
    },
    {
      "epoch": 0.4123116845671448,
      "grad_norm": 0.12244004249233213,
      "learning_rate": 0.00013265932316313013,
      "loss": 1.0153,
      "step": 5720
    },
    {
      "epoch": 0.4123837670294817,
      "grad_norm": 0.12302354160567353,
      "learning_rate": 0.00013263725536333723,
      "loss": 1.0663,
      "step": 5721
    },
    {
      "epoch": 0.4124558494918186,
      "grad_norm": 0.11539666092823268,
      "learning_rate": 0.0001326151857845219,
      "loss": 1.0518,
      "step": 5722
    },
    {
      "epoch": 0.41252793195415555,
      "grad_norm": 0.1419036649234459,
      "learning_rate": 0.00013259311442788716,
      "loss": 1.0655,
      "step": 5723
    },
    {
      "epoch": 0.41260001441649247,
      "grad_norm": 0.1849968987272934,
      "learning_rate": 0.00013257104129463614,
      "loss": 1.0457,
      "step": 5724
    },
    {
      "epoch": 0.4126720968788294,
      "grad_norm": 0.11948685130524682,
      "learning_rate": 0.00013254896638597194,
      "loss": 1.0757,
      "step": 5725
    },
    {
      "epoch": 0.4127441793411663,
      "grad_norm": 0.14631241691108857,
      "learning_rate": 0.00013252688970309792,
      "loss": 1.0101,
      "step": 5726
    },
    {
      "epoch": 0.4128162618035032,
      "grad_norm": 0.1583814194699134,
      "learning_rate": 0.00013250481124721736,
      "loss": 1.0581,
      "step": 5727
    },
    {
      "epoch": 0.41288834426584015,
      "grad_norm": 0.11479964276133894,
      "learning_rate": 0.00013248273101953387,
      "loss": 1.0184,
      "step": 5728
    },
    {
      "epoch": 0.412960426728177,
      "grad_norm": 0.1204057582987868,
      "learning_rate": 0.00013246064902125085,
      "loss": 1.1273,
      "step": 5729
    },
    {
      "epoch": 0.41303250919051393,
      "grad_norm": 0.19358187410303693,
      "learning_rate": 0.00013243856525357213,
      "loss": 1.1009,
      "step": 5730
    },
    {
      "epoch": 0.41310459165285085,
      "grad_norm": 0.1153399062028509,
      "learning_rate": 0.00013241647971770135,
      "loss": 1.0632,
      "step": 5731
    },
    {
      "epoch": 0.41317667411518777,
      "grad_norm": 0.12456247321807383,
      "learning_rate": 0.00013239439241484246,
      "loss": 1.1422,
      "step": 5732
    },
    {
      "epoch": 0.4132487565775247,
      "grad_norm": 0.10462174442765787,
      "learning_rate": 0.00013237230334619935,
      "loss": 1.0792,
      "step": 5733
    },
    {
      "epoch": 0.4133208390398616,
      "grad_norm": 0.202202973286419,
      "learning_rate": 0.0001323502125129761,
      "loss": 1.1315,
      "step": 5734
    },
    {
      "epoch": 0.41339292150219853,
      "grad_norm": 0.12776116326532072,
      "learning_rate": 0.00013232811991637685,
      "loss": 0.9465,
      "step": 5735
    },
    {
      "epoch": 0.41346500396453545,
      "grad_norm": 0.10747112744732495,
      "learning_rate": 0.00013230602555760587,
      "loss": 1.1283,
      "step": 5736
    },
    {
      "epoch": 0.41353708642687237,
      "grad_norm": 0.1294625405165921,
      "learning_rate": 0.00013228392943786748,
      "loss": 1.0353,
      "step": 5737
    },
    {
      "epoch": 0.41360916888920923,
      "grad_norm": 0.1016596220700341,
      "learning_rate": 0.00013226183155836608,
      "loss": 1.0666,
      "step": 5738
    },
    {
      "epoch": 0.41368125135154615,
      "grad_norm": 0.13127484122444386,
      "learning_rate": 0.00013223973192030624,
      "loss": 1.0468,
      "step": 5739
    },
    {
      "epoch": 0.4137533338138831,
      "grad_norm": 0.12020913558196396,
      "learning_rate": 0.00013221763052489262,
      "loss": 1.0593,
      "step": 5740
    },
    {
      "epoch": 0.41382541627622,
      "grad_norm": 0.40194857181148685,
      "learning_rate": 0.0001321955273733299,
      "loss": 1.0198,
      "step": 5741
    },
    {
      "epoch": 0.4138974987385569,
      "grad_norm": 0.10978725119956365,
      "learning_rate": 0.0001321734224668229,
      "loss": 0.9918,
      "step": 5742
    },
    {
      "epoch": 0.41396958120089383,
      "grad_norm": 0.12497876994716213,
      "learning_rate": 0.00013215131580657653,
      "loss": 1.0511,
      "step": 5743
    },
    {
      "epoch": 0.41404166366323075,
      "grad_norm": 0.11598135564886701,
      "learning_rate": 0.00013212920739379583,
      "loss": 1.0385,
      "step": 5744
    },
    {
      "epoch": 0.41411374612556767,
      "grad_norm": 0.14676231796840597,
      "learning_rate": 0.0001321070972296859,
      "loss": 1.0918,
      "step": 5745
    },
    {
      "epoch": 0.41418582858790454,
      "grad_norm": 0.17473591225036392,
      "learning_rate": 0.00013208498531545193,
      "loss": 1.0402,
      "step": 5746
    },
    {
      "epoch": 0.41425791105024146,
      "grad_norm": 0.1365951511517658,
      "learning_rate": 0.0001320628716522992,
      "loss": 1.0618,
      "step": 5747
    },
    {
      "epoch": 0.4143299935125784,
      "grad_norm": 0.340827656950944,
      "learning_rate": 0.00013204075624143315,
      "loss": 1.0745,
      "step": 5748
    },
    {
      "epoch": 0.4144020759749153,
      "grad_norm": 0.13049547759091604,
      "learning_rate": 0.00013201863908405921,
      "loss": 1.084,
      "step": 5749
    },
    {
      "epoch": 0.4144741584372522,
      "grad_norm": 1.3115146600416003,
      "learning_rate": 0.00013199652018138305,
      "loss": 1.1294,
      "step": 5750
    },
    {
      "epoch": 0.41454624089958914,
      "grad_norm": 3.3572295537309254,
      "learning_rate": 0.0001319743995346102,
      "loss": 1.0349,
      "step": 5751
    },
    {
      "epoch": 0.41461832336192606,
      "grad_norm": 0.15714353762488598,
      "learning_rate": 0.0001319522771449466,
      "loss": 1.0276,
      "step": 5752
    },
    {
      "epoch": 0.414690405824263,
      "grad_norm": 0.3141493095148568,
      "learning_rate": 0.000131930153013598,
      "loss": 1.1023,
      "step": 5753
    },
    {
      "epoch": 0.4147624882865999,
      "grad_norm": 0.2933420417197788,
      "learning_rate": 0.00013190802714177044,
      "loss": 1.1376,
      "step": 5754
    },
    {
      "epoch": 0.41483457074893676,
      "grad_norm": 0.17835598519631987,
      "learning_rate": 0.0001318858995306699,
      "loss": 1.0046,
      "step": 5755
    },
    {
      "epoch": 0.4149066532112737,
      "grad_norm": 0.12351810303027039,
      "learning_rate": 0.00013186377018150263,
      "loss": 1.1213,
      "step": 5756
    },
    {
      "epoch": 0.4149787356736106,
      "grad_norm": 0.14335456179949363,
      "learning_rate": 0.00013184163909547478,
      "loss": 1.1234,
      "step": 5757
    },
    {
      "epoch": 0.4150508181359475,
      "grad_norm": 0.2659032235988324,
      "learning_rate": 0.00013181950627379275,
      "loss": 1.0541,
      "step": 5758
    },
    {
      "epoch": 0.41512290059828444,
      "grad_norm": 0.11620751453945267,
      "learning_rate": 0.00013179737171766298,
      "loss": 1.0997,
      "step": 5759
    },
    {
      "epoch": 0.41519498306062136,
      "grad_norm": 0.4168488442571714,
      "learning_rate": 0.00013177523542829195,
      "loss": 1.0622,
      "step": 5760
    },
    {
      "epoch": 0.4152670655229583,
      "grad_norm": 0.13250850416816534,
      "learning_rate": 0.00013175309740688635,
      "loss": 1.04,
      "step": 5761
    },
    {
      "epoch": 0.4153391479852952,
      "grad_norm": 0.480393621299549,
      "learning_rate": 0.00013173095765465286,
      "loss": 1.1311,
      "step": 5762
    },
    {
      "epoch": 0.4154112304476321,
      "grad_norm": 0.33696257729856394,
      "learning_rate": 0.00013170881617279828,
      "loss": 1.0669,
      "step": 5763
    },
    {
      "epoch": 0.415483312909969,
      "grad_norm": 0.18194443230040877,
      "learning_rate": 0.0001316866729625296,
      "loss": 1.0816,
      "step": 5764
    },
    {
      "epoch": 0.4155553953723059,
      "grad_norm": 0.11865145966196292,
      "learning_rate": 0.0001316645280250537,
      "loss": 1.0465,
      "step": 5765
    },
    {
      "epoch": 0.4156274778346428,
      "grad_norm": 0.10782799316835107,
      "learning_rate": 0.00013164238136157777,
      "loss": 1.0523,
      "step": 5766
    },
    {
      "epoch": 0.41569956029697974,
      "grad_norm": 0.11662764318431353,
      "learning_rate": 0.00013162023297330898,
      "loss": 1.0792,
      "step": 5767
    },
    {
      "epoch": 0.41577164275931666,
      "grad_norm": 0.617873913855951,
      "learning_rate": 0.00013159808286145462,
      "loss": 1.2062,
      "step": 5768
    },
    {
      "epoch": 0.4158437252216536,
      "grad_norm": 0.48021546553074024,
      "learning_rate": 0.00013157593102722205,
      "loss": 1.0802,
      "step": 5769
    },
    {
      "epoch": 0.4159158076839905,
      "grad_norm": 0.2577018056933648,
      "learning_rate": 0.00013155377747181877,
      "loss": 1.1665,
      "step": 5770
    },
    {
      "epoch": 0.4159878901463274,
      "grad_norm": 0.1757919713292325,
      "learning_rate": 0.0001315316221964523,
      "loss": 1.068,
      "step": 5771
    },
    {
      "epoch": 0.4160599726086643,
      "grad_norm": 0.25896532591771687,
      "learning_rate": 0.00013150946520233033,
      "loss": 1.0626,
      "step": 5772
    },
    {
      "epoch": 0.4161320550710012,
      "grad_norm": 1.3132422754363184,
      "learning_rate": 0.00013148730649066063,
      "loss": 1.1242,
      "step": 5773
    },
    {
      "epoch": 0.4162041375333381,
      "grad_norm": 0.14443989494667991,
      "learning_rate": 0.00013146514606265103,
      "loss": 1.1302,
      "step": 5774
    },
    {
      "epoch": 0.41627621999567505,
      "grad_norm": 0.1702302779837644,
      "learning_rate": 0.00013144298391950949,
      "loss": 1.1224,
      "step": 5775
    },
    {
      "epoch": 0.41634830245801197,
      "grad_norm": 0.1603782169800543,
      "learning_rate": 0.00013142082006244402,
      "loss": 1.0621,
      "step": 5776
    },
    {
      "epoch": 0.4164203849203489,
      "grad_norm": 0.11547451565553117,
      "learning_rate": 0.00013139865449266272,
      "loss": 1.0621,
      "step": 5777
    },
    {
      "epoch": 0.4164924673826858,
      "grad_norm": 0.4359562556712193,
      "learning_rate": 0.0001313764872113739,
      "loss": 1.1061,
      "step": 5778
    },
    {
      "epoch": 0.4165645498450227,
      "grad_norm": 0.1012786711183534,
      "learning_rate": 0.00013135431821978578,
      "loss": 1.1214,
      "step": 5779
    },
    {
      "epoch": 0.41663663230735964,
      "grad_norm": 0.1379796900931547,
      "learning_rate": 0.00013133214751910686,
      "loss": 1.0644,
      "step": 5780
    },
    {
      "epoch": 0.4167087147696965,
      "grad_norm": 0.1116717043507401,
      "learning_rate": 0.00013130997511054559,
      "loss": 0.9934,
      "step": 5781
    },
    {
      "epoch": 0.41678079723203343,
      "grad_norm": 0.1833834373035439,
      "learning_rate": 0.00013128780099531056,
      "loss": 1.0503,
      "step": 5782
    },
    {
      "epoch": 0.41685287969437035,
      "grad_norm": 0.1178883323830432,
      "learning_rate": 0.00013126562517461044,
      "loss": 1.1358,
      "step": 5783
    },
    {
      "epoch": 0.41692496215670727,
      "grad_norm": 0.4183933682598301,
      "learning_rate": 0.00013124344764965407,
      "loss": 1.0065,
      "step": 5784
    },
    {
      "epoch": 0.4169970446190442,
      "grad_norm": 0.1183890783471969,
      "learning_rate": 0.00013122126842165027,
      "loss": 1.0118,
      "step": 5785
    },
    {
      "epoch": 0.4170691270813811,
      "grad_norm": 0.2014858223722653,
      "learning_rate": 0.00013119908749180802,
      "loss": 1.0688,
      "step": 5786
    },
    {
      "epoch": 0.41714120954371803,
      "grad_norm": 0.13128489967324392,
      "learning_rate": 0.00013117690486133646,
      "loss": 0.9946,
      "step": 5787
    },
    {
      "epoch": 0.41721329200605495,
      "grad_norm": 0.1731813974546806,
      "learning_rate": 0.00013115472053144457,
      "loss": 1.0783,
      "step": 5788
    },
    {
      "epoch": 0.41728537446839187,
      "grad_norm": 0.13846685648784757,
      "learning_rate": 0.00013113253450334176,
      "loss": 1.0609,
      "step": 5789
    },
    {
      "epoch": 0.41735745693072873,
      "grad_norm": 0.1484938116252876,
      "learning_rate": 0.00013111034677823727,
      "loss": 1.0276,
      "step": 5790
    },
    {
      "epoch": 0.41742953939306565,
      "grad_norm": 0.12256371598186545,
      "learning_rate": 0.00013108815735734058,
      "loss": 0.9851,
      "step": 5791
    },
    {
      "epoch": 0.41750162185540257,
      "grad_norm": 0.16094167524325528,
      "learning_rate": 0.00013106596624186116,
      "loss": 1.0599,
      "step": 5792
    },
    {
      "epoch": 0.4175737043177395,
      "grad_norm": 0.16207507523913184,
      "learning_rate": 0.00013104377343300868,
      "loss": 1.0633,
      "step": 5793
    },
    {
      "epoch": 0.4176457867800764,
      "grad_norm": 0.11483201945010948,
      "learning_rate": 0.0001310215789319928,
      "loss": 1.0598,
      "step": 5794
    },
    {
      "epoch": 0.41771786924241333,
      "grad_norm": 0.22792476146095508,
      "learning_rate": 0.00013099938274002337,
      "loss": 1.0381,
      "step": 5795
    },
    {
      "epoch": 0.41778995170475025,
      "grad_norm": 0.2572228246824519,
      "learning_rate": 0.00013097718485831022,
      "loss": 1.1409,
      "step": 5796
    },
    {
      "epoch": 0.41786203416708717,
      "grad_norm": 0.13766743495624176,
      "learning_rate": 0.00013095498528806337,
      "loss": 1.0297,
      "step": 5797
    },
    {
      "epoch": 0.41793411662942404,
      "grad_norm": 0.14397429982215235,
      "learning_rate": 0.00013093278403049293,
      "loss": 1.0801,
      "step": 5798
    },
    {
      "epoch": 0.41800619909176095,
      "grad_norm": 0.1165466224106999,
      "learning_rate": 0.000130910581086809,
      "loss": 1.0953,
      "step": 5799
    },
    {
      "epoch": 0.4180782815540979,
      "grad_norm": 0.19528306130538842,
      "learning_rate": 0.00013088837645822187,
      "loss": 1.0216,
      "step": 5800
    },
    {
      "epoch": 0.4181503640164348,
      "grad_norm": 0.2989218792636667,
      "learning_rate": 0.0001308661701459419,
      "loss": 1.0458,
      "step": 5801
    },
    {
      "epoch": 0.4182224464787717,
      "grad_norm": 0.11222413841589556,
      "learning_rate": 0.00013084396215117948,
      "loss": 1.1071,
      "step": 5802
    },
    {
      "epoch": 0.41829452894110863,
      "grad_norm": 0.16289015556341033,
      "learning_rate": 0.00013082175247514522,
      "loss": 1.1185,
      "step": 5803
    },
    {
      "epoch": 0.41836661140344555,
      "grad_norm": 0.11554687164446144,
      "learning_rate": 0.00013079954111904968,
      "loss": 1.0848,
      "step": 5804
    },
    {
      "epoch": 0.4184386938657825,
      "grad_norm": 0.13831462130697453,
      "learning_rate": 0.00013077732808410367,
      "loss": 1.0452,
      "step": 5805
    },
    {
      "epoch": 0.4185107763281194,
      "grad_norm": 0.432827740352779,
      "learning_rate": 0.00013075511337151788,
      "loss": 1.0665,
      "step": 5806
    },
    {
      "epoch": 0.41858285879045626,
      "grad_norm": 0.20068644534132163,
      "learning_rate": 0.0001307328969825033,
      "loss": 1.0403,
      "step": 5807
    },
    {
      "epoch": 0.4186549412527932,
      "grad_norm": 0.12632378635588945,
      "learning_rate": 0.00013071067891827086,
      "loss": 0.9919,
      "step": 5808
    },
    {
      "epoch": 0.4187270237151301,
      "grad_norm": 0.2659462782227974,
      "learning_rate": 0.0001306884591800317,
      "loss": 1.0452,
      "step": 5809
    },
    {
      "epoch": 0.418799106177467,
      "grad_norm": 0.3128790627227798,
      "learning_rate": 0.00013066623776899694,
      "loss": 1.0666,
      "step": 5810
    },
    {
      "epoch": 0.41887118863980394,
      "grad_norm": 0.11175084799813932,
      "learning_rate": 0.00013064401468637792,
      "loss": 1.0403,
      "step": 5811
    },
    {
      "epoch": 0.41894327110214086,
      "grad_norm": 0.10305990729585357,
      "learning_rate": 0.00013062178993338588,
      "loss": 1.0326,
      "step": 5812
    },
    {
      "epoch": 0.4190153535644778,
      "grad_norm": 0.16885058476149906,
      "learning_rate": 0.0001305995635112324,
      "loss": 1.0815,
      "step": 5813
    },
    {
      "epoch": 0.4190874360268147,
      "grad_norm": 0.1329225335476131,
      "learning_rate": 0.00013057733542112896,
      "loss": 1.0236,
      "step": 5814
    },
    {
      "epoch": 0.4191595184891516,
      "grad_norm": 0.22708183276897584,
      "learning_rate": 0.00013055510566428716,
      "loss": 1.0399,
      "step": 5815
    },
    {
      "epoch": 0.4192316009514885,
      "grad_norm": 0.1130002533788766,
      "learning_rate": 0.00013053287424191876,
      "loss": 1.0437,
      "step": 5816
    },
    {
      "epoch": 0.4193036834138254,
      "grad_norm": 0.16323846478470855,
      "learning_rate": 0.00013051064115523553,
      "loss": 1.0861,
      "step": 5817
    },
    {
      "epoch": 0.4193757658761623,
      "grad_norm": 0.19793212829460466,
      "learning_rate": 0.00013048840640544944,
      "loss": 1.1012,
      "step": 5818
    },
    {
      "epoch": 0.41944784833849924,
      "grad_norm": 0.17857000691930144,
      "learning_rate": 0.0001304661699937724,
      "loss": 1.1081,
      "step": 5819
    },
    {
      "epoch": 0.41951993080083616,
      "grad_norm": 0.11158072481518395,
      "learning_rate": 0.00013044393192141658,
      "loss": 1.0887,
      "step": 5820
    },
    {
      "epoch": 0.4195920132631731,
      "grad_norm": 0.14277926867643145,
      "learning_rate": 0.00013042169218959408,
      "loss": 1.1168,
      "step": 5821
    },
    {
      "epoch": 0.41966409572551,
      "grad_norm": 0.18062541541708652,
      "learning_rate": 0.00013039945079951722,
      "loss": 1.1272,
      "step": 5822
    },
    {
      "epoch": 0.4197361781878469,
      "grad_norm": 0.12437344084396817,
      "learning_rate": 0.00013037720775239834,
      "loss": 1.0636,
      "step": 5823
    },
    {
      "epoch": 0.4198082606501838,
      "grad_norm": 0.1208309144844748,
      "learning_rate": 0.00013035496304944982,
      "loss": 1.0877,
      "step": 5824
    },
    {
      "epoch": 0.4198803431125207,
      "grad_norm": 0.24429071859837306,
      "learning_rate": 0.00013033271669188432,
      "loss": 1.0262,
      "step": 5825
    },
    {
      "epoch": 0.4199524255748576,
      "grad_norm": 0.14233088665326976,
      "learning_rate": 0.00013031046868091435,
      "loss": 1.0696,
      "step": 5826
    },
    {
      "epoch": 0.42002450803719454,
      "grad_norm": 0.16483734874956388,
      "learning_rate": 0.0001302882190177527,
      "loss": 1.0121,
      "step": 5827
    },
    {
      "epoch": 0.42009659049953146,
      "grad_norm": 0.11252262463531379,
      "learning_rate": 0.0001302659677036121,
      "loss": 1.1401,
      "step": 5828
    },
    {
      "epoch": 0.4201686729618684,
      "grad_norm": 0.12700618708530048,
      "learning_rate": 0.00013024371473970554,
      "loss": 1.0335,
      "step": 5829
    },
    {
      "epoch": 0.4202407554242053,
      "grad_norm": 0.11394423045231597,
      "learning_rate": 0.00013022146012724593,
      "loss": 1.1213,
      "step": 5830
    },
    {
      "epoch": 0.4203128378865422,
      "grad_norm": 0.17316551130782354,
      "learning_rate": 0.0001301992038674464,
      "loss": 1.0659,
      "step": 5831
    },
    {
      "epoch": 0.42038492034887914,
      "grad_norm": 0.12074457882496992,
      "learning_rate": 0.00013017694596152005,
      "loss": 0.9963,
      "step": 5832
    },
    {
      "epoch": 0.420457002811216,
      "grad_norm": 0.1250100661004888,
      "learning_rate": 0.00013015468641068022,
      "loss": 1.1141,
      "step": 5833
    },
    {
      "epoch": 0.4205290852735529,
      "grad_norm": 0.1354892169876712,
      "learning_rate": 0.00013013242521614016,
      "loss": 1.019,
      "step": 5834
    },
    {
      "epoch": 0.42060116773588985,
      "grad_norm": 0.12164934226373586,
      "learning_rate": 0.00013011016237911342,
      "loss": 1.0328,
      "step": 5835
    },
    {
      "epoch": 0.42067325019822677,
      "grad_norm": 0.10956188155209319,
      "learning_rate": 0.0001300878979008134,
      "loss": 0.9669,
      "step": 5836
    },
    {
      "epoch": 0.4207453326605637,
      "grad_norm": 0.16956415133186048,
      "learning_rate": 0.00013006563178245378,
      "loss": 1.0755,
      "step": 5837
    },
    {
      "epoch": 0.4208174151229006,
      "grad_norm": 0.1586991563723089,
      "learning_rate": 0.00013004336402524827,
      "loss": 1.1011,
      "step": 5838
    },
    {
      "epoch": 0.4208894975852375,
      "grad_norm": 0.11436456602503965,
      "learning_rate": 0.00013002109463041065,
      "loss": 1.1095,
      "step": 5839
    },
    {
      "epoch": 0.42096158004757445,
      "grad_norm": 0.14391311726790773,
      "learning_rate": 0.0001299988235991548,
      "loss": 1.0449,
      "step": 5840
    },
    {
      "epoch": 0.42103366250991137,
      "grad_norm": 0.12373803632579662,
      "learning_rate": 0.00012997655093269466,
      "loss": 1.0671,
      "step": 5841
    },
    {
      "epoch": 0.42110574497224823,
      "grad_norm": 0.15039637988148757,
      "learning_rate": 0.00012995427663224432,
      "loss": 1.0477,
      "step": 5842
    },
    {
      "epoch": 0.42117782743458515,
      "grad_norm": 0.11618347116863106,
      "learning_rate": 0.00012993200069901794,
      "loss": 1.0534,
      "step": 5843
    },
    {
      "epoch": 0.42124990989692207,
      "grad_norm": 0.1141674332109756,
      "learning_rate": 0.00012990972313422973,
      "loss": 0.9563,
      "step": 5844
    },
    {
      "epoch": 0.421321992359259,
      "grad_norm": 0.1067988987075522,
      "learning_rate": 0.00012988744393909403,
      "loss": 1.0229,
      "step": 5845
    },
    {
      "epoch": 0.4213940748215959,
      "grad_norm": 0.3217865126193961,
      "learning_rate": 0.00012986516311482525,
      "loss": 1.0705,
      "step": 5846
    },
    {
      "epoch": 0.42146615728393283,
      "grad_norm": 0.11155877552211514,
      "learning_rate": 0.00012984288066263792,
      "loss": 1.0397,
      "step": 5847
    },
    {
      "epoch": 0.42153823974626975,
      "grad_norm": 0.11118108999866563,
      "learning_rate": 0.0001298205965837466,
      "loss": 1.0852,
      "step": 5848
    },
    {
      "epoch": 0.42161032220860667,
      "grad_norm": 0.15013671776130322,
      "learning_rate": 0.00012979831087936598,
      "loss": 1.1355,
      "step": 5849
    },
    {
      "epoch": 0.42168240467094353,
      "grad_norm": 0.14938127765083603,
      "learning_rate": 0.00012977602355071085,
      "loss": 1.1092,
      "step": 5850
    },
    {
      "epoch": 0.42175448713328045,
      "grad_norm": 0.11972799706701055,
      "learning_rate": 0.00012975373459899603,
      "loss": 1.1124,
      "step": 5851
    },
    {
      "epoch": 0.4218265695956174,
      "grad_norm": 0.1531773015458602,
      "learning_rate": 0.0001297314440254365,
      "loss": 1.0851,
      "step": 5852
    },
    {
      "epoch": 0.4218986520579543,
      "grad_norm": 0.13263834326957158,
      "learning_rate": 0.00012970915183124727,
      "loss": 1.0613,
      "step": 5853
    },
    {
      "epoch": 0.4219707345202912,
      "grad_norm": 0.1437248462128305,
      "learning_rate": 0.00012968685801764348,
      "loss": 1.1628,
      "step": 5854
    },
    {
      "epoch": 0.42204281698262813,
      "grad_norm": 0.13257756351331768,
      "learning_rate": 0.00012966456258584036,
      "loss": 1.0899,
      "step": 5855
    },
    {
      "epoch": 0.42211489944496505,
      "grad_norm": 0.3000154556894525,
      "learning_rate": 0.0001296422655370532,
      "loss": 1.1212,
      "step": 5856
    },
    {
      "epoch": 0.422186981907302,
      "grad_norm": 0.12299771351085323,
      "learning_rate": 0.00012961996687249738,
      "loss": 1.024,
      "step": 5857
    },
    {
      "epoch": 0.4222590643696389,
      "grad_norm": 0.10930847489280292,
      "learning_rate": 0.00012959766659338836,
      "loss": 1.0589,
      "step": 5858
    },
    {
      "epoch": 0.42233114683197576,
      "grad_norm": 0.19460623760403414,
      "learning_rate": 0.00012957536470094172,
      "loss": 1.1554,
      "step": 5859
    },
    {
      "epoch": 0.4224032292943127,
      "grad_norm": 0.1351158010716234,
      "learning_rate": 0.00012955306119637316,
      "loss": 1.061,
      "step": 5860
    },
    {
      "epoch": 0.4224753117566496,
      "grad_norm": 0.11833187756817472,
      "learning_rate": 0.00012953075608089832,
      "loss": 1.0608,
      "step": 5861
    },
    {
      "epoch": 0.4225473942189865,
      "grad_norm": 0.19144523482156117,
      "learning_rate": 0.00012950844935573312,
      "loss": 1.0136,
      "step": 5862
    },
    {
      "epoch": 0.42261947668132344,
      "grad_norm": 0.12326833000450221,
      "learning_rate": 0.0001294861410220934,
      "loss": 0.9864,
      "step": 5863
    },
    {
      "epoch": 0.42269155914366036,
      "grad_norm": 0.6114238582927436,
      "learning_rate": 0.00012946383108119525,
      "loss": 1.0883,
      "step": 5864
    },
    {
      "epoch": 0.4227636416059973,
      "grad_norm": 0.15194536131013367,
      "learning_rate": 0.00012944151953425464,
      "loss": 1.0258,
      "step": 5865
    },
    {
      "epoch": 0.4228357240683342,
      "grad_norm": 0.12853299551128738,
      "learning_rate": 0.00012941920638248792,
      "loss": 1.0253,
      "step": 5866
    },
    {
      "epoch": 0.4229078065306711,
      "grad_norm": 0.11508682548556659,
      "learning_rate": 0.0001293968916271112,
      "loss": 1.1139,
      "step": 5867
    },
    {
      "epoch": 0.422979888993008,
      "grad_norm": 0.13875480730738227,
      "learning_rate": 0.00012937457526934092,
      "loss": 1.0592,
      "step": 5868
    },
    {
      "epoch": 0.4230519714553449,
      "grad_norm": 0.128665538584716,
      "learning_rate": 0.00012935225731039348,
      "loss": 1.0686,
      "step": 5869
    },
    {
      "epoch": 0.4231240539176818,
      "grad_norm": 0.14604638038571696,
      "learning_rate": 0.00012932993775148543,
      "loss": 0.9745,
      "step": 5870
    },
    {
      "epoch": 0.42319613638001874,
      "grad_norm": 0.13895254705253438,
      "learning_rate": 0.0001293076165938334,
      "loss": 1.0329,
      "step": 5871
    },
    {
      "epoch": 0.42326821884235566,
      "grad_norm": 0.22451725799864738,
      "learning_rate": 0.00012928529383865404,
      "loss": 1.0677,
      "step": 5872
    },
    {
      "epoch": 0.4233403013046926,
      "grad_norm": 0.1426751666685977,
      "learning_rate": 0.00012926296948716418,
      "loss": 1.005,
      "step": 5873
    },
    {
      "epoch": 0.4234123837670295,
      "grad_norm": 0.14409653831013855,
      "learning_rate": 0.0001292406435405807,
      "loss": 1.0549,
      "step": 5874
    },
    {
      "epoch": 0.4234844662293664,
      "grad_norm": 0.12075925354700809,
      "learning_rate": 0.00012921831600012053,
      "loss": 1.1278,
      "step": 5875
    },
    {
      "epoch": 0.4235565486917033,
      "grad_norm": 0.11303280482123183,
      "learning_rate": 0.00012919598686700077,
      "loss": 1.0662,
      "step": 5876
    },
    {
      "epoch": 0.4236286311540402,
      "grad_norm": 0.16501336918379197,
      "learning_rate": 0.0001291736561424385,
      "loss": 1.0624,
      "step": 5877
    },
    {
      "epoch": 0.4237007136163771,
      "grad_norm": 0.10547805449398051,
      "learning_rate": 0.000129151323827651,
      "loss": 1.0116,
      "step": 5878
    },
    {
      "epoch": 0.42377279607871404,
      "grad_norm": 0.10763573346624317,
      "learning_rate": 0.00012912898992385556,
      "loss": 0.9775,
      "step": 5879
    },
    {
      "epoch": 0.42384487854105096,
      "grad_norm": 0.12954622940626667,
      "learning_rate": 0.00012910665443226959,
      "loss": 1.043,
      "step": 5880
    },
    {
      "epoch": 0.4239169610033879,
      "grad_norm": 0.16263607682084655,
      "learning_rate": 0.0001290843173541105,
      "loss": 1.0086,
      "step": 5881
    },
    {
      "epoch": 0.4239890434657248,
      "grad_norm": 0.1850103307330114,
      "learning_rate": 0.00012906197869059596,
      "loss": 1.0317,
      "step": 5882
    },
    {
      "epoch": 0.4240611259280617,
      "grad_norm": 0.11998995246093407,
      "learning_rate": 0.00012903963844294357,
      "loss": 1.0967,
      "step": 5883
    },
    {
      "epoch": 0.42413320839039864,
      "grad_norm": 0.14102604058911633,
      "learning_rate": 0.00012901729661237113,
      "loss": 1.1242,
      "step": 5884
    },
    {
      "epoch": 0.4242052908527355,
      "grad_norm": 0.1474631164703126,
      "learning_rate": 0.00012899495320009636,
      "loss": 1.1136,
      "step": 5885
    },
    {
      "epoch": 0.4242773733150724,
      "grad_norm": 0.1885190761773702,
      "learning_rate": 0.0001289726082073373,
      "loss": 1.0503,
      "step": 5886
    },
    {
      "epoch": 0.42434945577740935,
      "grad_norm": 0.13098698033354506,
      "learning_rate": 0.00012895026163531185,
      "loss": 0.985,
      "step": 5887
    },
    {
      "epoch": 0.42442153823974627,
      "grad_norm": 0.12039569839336427,
      "learning_rate": 0.00012892791348523817,
      "loss": 1.1198,
      "step": 5888
    },
    {
      "epoch": 0.4244936207020832,
      "grad_norm": 0.11843078378191792,
      "learning_rate": 0.00012890556375833444,
      "loss": 1.029,
      "step": 5889
    },
    {
      "epoch": 0.4245657031644201,
      "grad_norm": 0.11598918617465045,
      "learning_rate": 0.00012888321245581885,
      "loss": 1.0478,
      "step": 5890
    },
    {
      "epoch": 0.424637785626757,
      "grad_norm": 0.20460336006464871,
      "learning_rate": 0.0001288608595789098,
      "loss": 1.0324,
      "step": 5891
    },
    {
      "epoch": 0.42470986808909394,
      "grad_norm": 0.19634724077481586,
      "learning_rate": 0.0001288385051288257,
      "loss": 1.0441,
      "step": 5892
    },
    {
      "epoch": 0.4247819505514308,
      "grad_norm": 0.11784200001395477,
      "learning_rate": 0.00012881614910678505,
      "loss": 1.1227,
      "step": 5893
    },
    {
      "epoch": 0.42485403301376773,
      "grad_norm": 0.1627096136154011,
      "learning_rate": 0.0001287937915140065,
      "loss": 1.0634,
      "step": 5894
    },
    {
      "epoch": 0.42492611547610465,
      "grad_norm": 0.1119371649813452,
      "learning_rate": 0.00012877143235170873,
      "loss": 1.0805,
      "step": 5895
    },
    {
      "epoch": 0.42499819793844157,
      "grad_norm": 0.12192155082969519,
      "learning_rate": 0.00012874907162111046,
      "loss": 1.0099,
      "step": 5896
    },
    {
      "epoch": 0.4250702804007785,
      "grad_norm": 0.15008868883805676,
      "learning_rate": 0.00012872670932343063,
      "loss": 1.0176,
      "step": 5897
    },
    {
      "epoch": 0.4251423628631154,
      "grad_norm": 0.13997908186542835,
      "learning_rate": 0.00012870434545988812,
      "loss": 1.0482,
      "step": 5898
    },
    {
      "epoch": 0.42521444532545233,
      "grad_norm": 0.12411349946284438,
      "learning_rate": 0.000128681980031702,
      "loss": 1.1219,
      "step": 5899
    },
    {
      "epoch": 0.42528652778778925,
      "grad_norm": 0.12028301395808204,
      "learning_rate": 0.0001286596130400914,
      "loss": 1.0824,
      "step": 5900
    },
    {
      "epoch": 0.42535861025012617,
      "grad_norm": 0.13342905308274114,
      "learning_rate": 0.00012863724448627543,
      "loss": 1.0619,
      "step": 5901
    },
    {
      "epoch": 0.42543069271246303,
      "grad_norm": 0.22890336564081212,
      "learning_rate": 0.0001286148743714735,
      "loss": 1.0401,
      "step": 5902
    },
    {
      "epoch": 0.42550277517479995,
      "grad_norm": 0.12423018552658781,
      "learning_rate": 0.00012859250269690487,
      "loss": 0.9458,
      "step": 5903
    },
    {
      "epoch": 0.42557485763713687,
      "grad_norm": 0.14769696178337255,
      "learning_rate": 0.00012857012946378908,
      "loss": 1.0431,
      "step": 5904
    },
    {
      "epoch": 0.4256469400994738,
      "grad_norm": 0.1444277789661778,
      "learning_rate": 0.0001285477546733456,
      "loss": 1.0575,
      "step": 5905
    },
    {
      "epoch": 0.4257190225618107,
      "grad_norm": 0.11959593353337149,
      "learning_rate": 0.00012852537832679415,
      "loss": 1.1083,
      "step": 5906
    },
    {
      "epoch": 0.42579110502414763,
      "grad_norm": 0.12149654927156935,
      "learning_rate": 0.00012850300042535433,
      "loss": 1.0367,
      "step": 5907
    },
    {
      "epoch": 0.42586318748648455,
      "grad_norm": 0.1948985938245956,
      "learning_rate": 0.00012848062097024605,
      "loss": 1.0883,
      "step": 5908
    },
    {
      "epoch": 0.42593526994882147,
      "grad_norm": 0.13591317837507574,
      "learning_rate": 0.00012845823996268906,
      "loss": 1.0576,
      "step": 5909
    },
    {
      "epoch": 0.4260073524111584,
      "grad_norm": 0.2129790963706379,
      "learning_rate": 0.00012843585740390348,
      "loss": 1.0201,
      "step": 5910
    },
    {
      "epoch": 0.42607943487349526,
      "grad_norm": 0.21434191350287624,
      "learning_rate": 0.00012841347329510923,
      "loss": 1.0631,
      "step": 5911
    },
    {
      "epoch": 0.4261515173358322,
      "grad_norm": 0.10151162728503786,
      "learning_rate": 0.0001283910876375265,
      "loss": 1.0381,
      "step": 5912
    },
    {
      "epoch": 0.4262235997981691,
      "grad_norm": 0.2190244359423758,
      "learning_rate": 0.00012836870043237547,
      "loss": 1.069,
      "step": 5913
    },
    {
      "epoch": 0.426295682260506,
      "grad_norm": 0.10547828213391817,
      "learning_rate": 0.00012834631168087652,
      "loss": 1.0545,
      "step": 5914
    },
    {
      "epoch": 0.42636776472284293,
      "grad_norm": 0.12504236927514434,
      "learning_rate": 0.00012832392138424997,
      "loss": 1.0505,
      "step": 5915
    },
    {
      "epoch": 0.42643984718517985,
      "grad_norm": 0.19961902839286044,
      "learning_rate": 0.0001283015295437163,
      "loss": 1.081,
      "step": 5916
    },
    {
      "epoch": 0.4265119296475168,
      "grad_norm": 0.206210879315371,
      "learning_rate": 0.0001282791361604961,
      "loss": 1.0669,
      "step": 5917
    },
    {
      "epoch": 0.4265840121098537,
      "grad_norm": 0.12245937296618183,
      "learning_rate": 0.00012825674123580997,
      "loss": 1.1194,
      "step": 5918
    },
    {
      "epoch": 0.42665609457219056,
      "grad_norm": 0.1105767060907629,
      "learning_rate": 0.00012823434477087867,
      "loss": 1.0423,
      "step": 5919
    },
    {
      "epoch": 0.4267281770345275,
      "grad_norm": 0.1915853542650012,
      "learning_rate": 0.00012821194676692295,
      "loss": 1.0456,
      "step": 5920
    },
    {
      "epoch": 0.4268002594968644,
      "grad_norm": 0.1429880362624331,
      "learning_rate": 0.00012818954722516377,
      "loss": 1.0067,
      "step": 5921
    },
    {
      "epoch": 0.4268723419592013,
      "grad_norm": 0.12748662182025206,
      "learning_rate": 0.00012816714614682207,
      "loss": 1.0143,
      "step": 5922
    },
    {
      "epoch": 0.42694442442153824,
      "grad_norm": 0.651170695815434,
      "learning_rate": 0.00012814474353311893,
      "loss": 1.1137,
      "step": 5923
    },
    {
      "epoch": 0.42701650688387516,
      "grad_norm": 0.15844324083709976,
      "learning_rate": 0.00012812233938527543,
      "loss": 0.9464,
      "step": 5924
    },
    {
      "epoch": 0.4270885893462121,
      "grad_norm": 0.11715221171234712,
      "learning_rate": 0.00012809993370451288,
      "loss": 1.0234,
      "step": 5925
    },
    {
      "epoch": 0.427160671808549,
      "grad_norm": 0.16193224771604764,
      "learning_rate": 0.00012807752649205255,
      "loss": 1.051,
      "step": 5926
    },
    {
      "epoch": 0.4272327542708859,
      "grad_norm": 0.15431015218144722,
      "learning_rate": 0.00012805511774911584,
      "loss": 1.132,
      "step": 5927
    },
    {
      "epoch": 0.4273048367332228,
      "grad_norm": 0.12957554586663417,
      "learning_rate": 0.00012803270747692422,
      "loss": 1.0465,
      "step": 5928
    },
    {
      "epoch": 0.4273769191955597,
      "grad_norm": 0.142406222828008,
      "learning_rate": 0.00012801029567669926,
      "loss": 1.0454,
      "step": 5929
    },
    {
      "epoch": 0.4274490016578966,
      "grad_norm": 0.13290079045266667,
      "learning_rate": 0.00012798788234966258,
      "loss": 1.1024,
      "step": 5930
    },
    {
      "epoch": 0.42752108412023354,
      "grad_norm": 0.13239511806950777,
      "learning_rate": 0.00012796546749703594,
      "loss": 1.0678,
      "step": 5931
    },
    {
      "epoch": 0.42759316658257046,
      "grad_norm": 0.14263043900266956,
      "learning_rate": 0.00012794305112004113,
      "loss": 1.0705,
      "step": 5932
    },
    {
      "epoch": 0.4276652490449074,
      "grad_norm": 0.11843516286713329,
      "learning_rate": 0.00012792063321990008,
      "loss": 1.0482,
      "step": 5933
    },
    {
      "epoch": 0.4277373315072443,
      "grad_norm": 0.11678439028407127,
      "learning_rate": 0.00012789821379783468,
      "loss": 1.0082,
      "step": 5934
    },
    {
      "epoch": 0.4278094139695812,
      "grad_norm": 0.10489230950743197,
      "learning_rate": 0.0001278757928550671,
      "loss": 1.0091,
      "step": 5935
    },
    {
      "epoch": 0.42788149643191814,
      "grad_norm": 0.15540075247678137,
      "learning_rate": 0.00012785337039281937,
      "loss": 1.1033,
      "step": 5936
    },
    {
      "epoch": 0.427953578894255,
      "grad_norm": 0.11940433714717634,
      "learning_rate": 0.00012783094641231383,
      "loss": 0.9698,
      "step": 5937
    },
    {
      "epoch": 0.4280256613565919,
      "grad_norm": 0.13604457955083044,
      "learning_rate": 0.00012780852091477267,
      "loss": 1.0415,
      "step": 5938
    },
    {
      "epoch": 0.42809774381892884,
      "grad_norm": 0.3976295368286561,
      "learning_rate": 0.0001277860939014184,
      "loss": 1.1059,
      "step": 5939
    },
    {
      "epoch": 0.42816982628126576,
      "grad_norm": 0.11588610851851637,
      "learning_rate": 0.00012776366537347336,
      "loss": 1.0459,
      "step": 5940
    },
    {
      "epoch": 0.4282419087436027,
      "grad_norm": 0.12319795590984174,
      "learning_rate": 0.00012774123533216023,
      "loss": 1.0255,
      "step": 5941
    },
    {
      "epoch": 0.4283139912059396,
      "grad_norm": 0.11663380962436176,
      "learning_rate": 0.00012771880377870156,
      "loss": 1.0556,
      "step": 5942
    },
    {
      "epoch": 0.4283860736682765,
      "grad_norm": 0.13478892752036165,
      "learning_rate": 0.00012769637071432011,
      "loss": 1.1098,
      "step": 5943
    },
    {
      "epoch": 0.42845815613061344,
      "grad_norm": 0.32297012903984057,
      "learning_rate": 0.0001276739361402387,
      "loss": 1.0105,
      "step": 5944
    },
    {
      "epoch": 0.4285302385929503,
      "grad_norm": 0.12462451682253319,
      "learning_rate": 0.00012765150005768016,
      "loss": 1.0319,
      "step": 5945
    },
    {
      "epoch": 0.4286023210552872,
      "grad_norm": 0.11619887663646579,
      "learning_rate": 0.00012762906246786753,
      "loss": 1.0539,
      "step": 5946
    },
    {
      "epoch": 0.42867440351762415,
      "grad_norm": 0.1730233936800411,
      "learning_rate": 0.00012760662337202378,
      "loss": 1.0361,
      "step": 5947
    },
    {
      "epoch": 0.42874648597996107,
      "grad_norm": 0.1134047578567967,
      "learning_rate": 0.0001275841827713721,
      "loss": 0.9128,
      "step": 5948
    },
    {
      "epoch": 0.428818568442298,
      "grad_norm": 0.11834547251261994,
      "learning_rate": 0.0001275617406671357,
      "loss": 1.0622,
      "step": 5949
    },
    {
      "epoch": 0.4288906509046349,
      "grad_norm": 0.18075578007704365,
      "learning_rate": 0.00012753929706053782,
      "loss": 1.0764,
      "step": 5950
    },
    {
      "epoch": 0.4289627333669718,
      "grad_norm": 0.1783716777066071,
      "learning_rate": 0.0001275168519528019,
      "loss": 1.0439,
      "step": 5951
    },
    {
      "epoch": 0.42903481582930875,
      "grad_norm": 0.11850031278442451,
      "learning_rate": 0.00012749440534515135,
      "loss": 1.0493,
      "step": 5952
    },
    {
      "epoch": 0.42910689829164567,
      "grad_norm": 0.41385232772900366,
      "learning_rate": 0.00012747195723880977,
      "loss": 1.0318,
      "step": 5953
    },
    {
      "epoch": 0.42917898075398253,
      "grad_norm": 0.1153084186308926,
      "learning_rate": 0.00012744950763500073,
      "loss": 1.0455,
      "step": 5954
    },
    {
      "epoch": 0.42925106321631945,
      "grad_norm": 0.16604197011385521,
      "learning_rate": 0.000127427056534948,
      "loss": 1.0538,
      "step": 5955
    },
    {
      "epoch": 0.42932314567865637,
      "grad_norm": 0.14974184876505792,
      "learning_rate": 0.00012740460393987526,
      "loss": 1.1223,
      "step": 5956
    },
    {
      "epoch": 0.4293952281409933,
      "grad_norm": 0.13173176067585937,
      "learning_rate": 0.00012738214985100646,
      "loss": 1.0473,
      "step": 5957
    },
    {
      "epoch": 0.4294673106033302,
      "grad_norm": 0.11445356094288789,
      "learning_rate": 0.00012735969426956551,
      "loss": 1.0565,
      "step": 5958
    },
    {
      "epoch": 0.42953939306566713,
      "grad_norm": 0.12070281821323858,
      "learning_rate": 0.00012733723719677648,
      "loss": 1.1042,
      "step": 5959
    },
    {
      "epoch": 0.42961147552800405,
      "grad_norm": 0.10321160212808458,
      "learning_rate": 0.00012731477863386342,
      "loss": 1.053,
      "step": 5960
    },
    {
      "epoch": 0.42968355799034097,
      "grad_norm": 0.24726741514126702,
      "learning_rate": 0.00012729231858205062,
      "loss": 0.978,
      "step": 5961
    },
    {
      "epoch": 0.4297556404526779,
      "grad_norm": 0.25956325099743494,
      "learning_rate": 0.00012726985704256222,
      "loss": 1.0007,
      "step": 5962
    },
    {
      "epoch": 0.42982772291501475,
      "grad_norm": 0.1269626527746557,
      "learning_rate": 0.00012724739401662272,
      "loss": 1.0075,
      "step": 5963
    },
    {
      "epoch": 0.4298998053773517,
      "grad_norm": 0.13018377383930008,
      "learning_rate": 0.00012722492950545644,
      "loss": 0.9801,
      "step": 5964
    },
    {
      "epoch": 0.4299718878396886,
      "grad_norm": 0.1155578354880112,
      "learning_rate": 0.00012720246351028794,
      "loss": 1.0299,
      "step": 5965
    },
    {
      "epoch": 0.4300439703020255,
      "grad_norm": 0.11168215435362061,
      "learning_rate": 0.00012717999603234183,
      "loss": 1.067,
      "step": 5966
    },
    {
      "epoch": 0.43011605276436243,
      "grad_norm": 0.12494287733535507,
      "learning_rate": 0.00012715752707284278,
      "loss": 0.9826,
      "step": 5967
    },
    {
      "epoch": 0.43018813522669935,
      "grad_norm": 0.1633530273677273,
      "learning_rate": 0.00012713505663301556,
      "loss": 1.0563,
      "step": 5968
    },
    {
      "epoch": 0.4302602176890363,
      "grad_norm": 0.10746936667122119,
      "learning_rate": 0.00012711258471408496,
      "loss": 1.1115,
      "step": 5969
    },
    {
      "epoch": 0.4303323001513732,
      "grad_norm": 0.11808859304730335,
      "learning_rate": 0.00012709011131727596,
      "loss": 1.1103,
      "step": 5970
    },
    {
      "epoch": 0.43040438261371006,
      "grad_norm": 0.5489044459128708,
      "learning_rate": 0.00012706763644381355,
      "loss": 1.0154,
      "step": 5971
    },
    {
      "epoch": 0.430476465076047,
      "grad_norm": 0.1073807426661013,
      "learning_rate": 0.00012704516009492278,
      "loss": 1.0562,
      "step": 5972
    },
    {
      "epoch": 0.4305485475383839,
      "grad_norm": 0.13665263971883712,
      "learning_rate": 0.00012702268227182884,
      "loss": 0.9522,
      "step": 5973
    },
    {
      "epoch": 0.4306206300007208,
      "grad_norm": 0.1106340687154567,
      "learning_rate": 0.00012700020297575694,
      "loss": 1.052,
      "step": 5974
    },
    {
      "epoch": 0.43069271246305774,
      "grad_norm": 0.12025655947086794,
      "learning_rate": 0.00012697772220793244,
      "loss": 1.0283,
      "step": 5975
    },
    {
      "epoch": 0.43076479492539466,
      "grad_norm": 0.10433599818461824,
      "learning_rate": 0.00012695523996958074,
      "loss": 1.012,
      "step": 5976
    },
    {
      "epoch": 0.4308368773877316,
      "grad_norm": 0.16702321381973514,
      "learning_rate": 0.00012693275626192733,
      "loss": 1.0322,
      "step": 5977
    },
    {
      "epoch": 0.4309089598500685,
      "grad_norm": 0.1399930326184591,
      "learning_rate": 0.00012691027108619772,
      "loss": 1.0759,
      "step": 5978
    },
    {
      "epoch": 0.4309810423124054,
      "grad_norm": 0.34112035527548784,
      "learning_rate": 0.00012688778444361765,
      "loss": 1.114,
      "step": 5979
    },
    {
      "epoch": 0.4310531247747423,
      "grad_norm": 0.1666145248880734,
      "learning_rate": 0.0001268652963354127,
      "loss": 1.0583,
      "step": 5980
    },
    {
      "epoch": 0.4311252072370792,
      "grad_norm": 0.13537350363733286,
      "learning_rate": 0.00012684280676280882,
      "loss": 0.9846,
      "step": 5981
    },
    {
      "epoch": 0.4311972896994161,
      "grad_norm": 0.11515210089409138,
      "learning_rate": 0.00012682031572703177,
      "loss": 1.0167,
      "step": 5982
    },
    {
      "epoch": 0.43126937216175304,
      "grad_norm": 0.0994741210812277,
      "learning_rate": 0.0001267978232293076,
      "loss": 0.994,
      "step": 5983
    },
    {
      "epoch": 0.43134145462408996,
      "grad_norm": 0.2168827910239142,
      "learning_rate": 0.00012677532927086234,
      "loss": 1.0195,
      "step": 5984
    },
    {
      "epoch": 0.4314135370864269,
      "grad_norm": 0.1226196422769662,
      "learning_rate": 0.00012675283385292212,
      "loss": 1.051,
      "step": 5985
    },
    {
      "epoch": 0.4314856195487638,
      "grad_norm": 0.16928338934953358,
      "learning_rate": 0.00012673033697671303,
      "loss": 1.0162,
      "step": 5986
    },
    {
      "epoch": 0.4315577020111007,
      "grad_norm": 0.13000424685794415,
      "learning_rate": 0.00012670783864346153,
      "loss": 1.0843,
      "step": 5987
    },
    {
      "epoch": 0.43162978447343764,
      "grad_norm": 0.12863985947508574,
      "learning_rate": 0.00012668533885439384,
      "loss": 1.0631,
      "step": 5988
    },
    {
      "epoch": 0.4317018669357745,
      "grad_norm": 0.11366324898484315,
      "learning_rate": 0.00012666283761073648,
      "loss": 1.009,
      "step": 5989
    },
    {
      "epoch": 0.4317739493981114,
      "grad_norm": 0.12103171209764435,
      "learning_rate": 0.0001266403349137159,
      "loss": 1.0283,
      "step": 5990
    },
    {
      "epoch": 0.43184603186044834,
      "grad_norm": 0.12285418105524938,
      "learning_rate": 0.00012661783076455875,
      "loss": 1.0682,
      "step": 5991
    },
    {
      "epoch": 0.43191811432278526,
      "grad_norm": 0.1544584785194945,
      "learning_rate": 0.00012659532516449168,
      "loss": 1.0304,
      "step": 5992
    },
    {
      "epoch": 0.4319901967851222,
      "grad_norm": 0.1374579401379385,
      "learning_rate": 0.0001265728181147415,
      "loss": 1.0556,
      "step": 5993
    },
    {
      "epoch": 0.4320622792474591,
      "grad_norm": 0.2611702065707183,
      "learning_rate": 0.00012655030961653498,
      "loss": 1.0352,
      "step": 5994
    },
    {
      "epoch": 0.432134361709796,
      "grad_norm": 0.11604494937847779,
      "learning_rate": 0.00012652779967109904,
      "loss": 1.0715,
      "step": 5995
    },
    {
      "epoch": 0.43220644417213294,
      "grad_norm": 0.6468869898932671,
      "learning_rate": 0.00012650528827966073,
      "loss": 1.0536,
      "step": 5996
    },
    {
      "epoch": 0.4322785266344698,
      "grad_norm": 0.11410483850646093,
      "learning_rate": 0.00012648277544344706,
      "loss": 1.0904,
      "step": 5997
    },
    {
      "epoch": 0.4323506090968067,
      "grad_norm": 0.10636883959231606,
      "learning_rate": 0.00012646026116368522,
      "loss": 1.009,
      "step": 5998
    },
    {
      "epoch": 0.43242269155914365,
      "grad_norm": 0.14337697062131763,
      "learning_rate": 0.0001264377454416024,
      "loss": 1.0681,
      "step": 5999
    },
    {
      "epoch": 0.43249477402148057,
      "grad_norm": 0.21184437191382824,
      "learning_rate": 0.00012641522827842598,
      "loss": 1.0592,
      "step": 6000
    },
    {
      "epoch": 0.4325668564838175,
      "grad_norm": 0.40662271530639,
      "learning_rate": 0.00012639270967538328,
      "loss": 1.1413,
      "step": 6001
    },
    {
      "epoch": 0.4326389389461544,
      "grad_norm": 0.13218576394153347,
      "learning_rate": 0.00012637018963370178,
      "loss": 1.0621,
      "step": 6002
    },
    {
      "epoch": 0.4327110214084913,
      "grad_norm": 0.12420345279887555,
      "learning_rate": 0.00012634766815460902,
      "loss": 1.0329,
      "step": 6003
    },
    {
      "epoch": 0.43278310387082825,
      "grad_norm": 0.1280796587227618,
      "learning_rate": 0.00012632514523933265,
      "loss": 1.0346,
      "step": 6004
    },
    {
      "epoch": 0.43285518633316517,
      "grad_norm": 0.126247167566334,
      "learning_rate": 0.00012630262088910037,
      "loss": 1.0452,
      "step": 6005
    },
    {
      "epoch": 0.43292726879550203,
      "grad_norm": 0.1395387843743815,
      "learning_rate": 0.00012628009510513993,
      "loss": 1.0344,
      "step": 6006
    },
    {
      "epoch": 0.43299935125783895,
      "grad_norm": 0.12193426088189889,
      "learning_rate": 0.00012625756788867918,
      "loss": 1.0412,
      "step": 6007
    },
    {
      "epoch": 0.43307143372017587,
      "grad_norm": 1.128788344334269,
      "learning_rate": 0.0001262350392409461,
      "loss": 1.0715,
      "step": 6008
    },
    {
      "epoch": 0.4331435161825128,
      "grad_norm": 0.10509622015035273,
      "learning_rate": 0.00012621250916316864,
      "loss": 0.9548,
      "step": 6009
    },
    {
      "epoch": 0.4332155986448497,
      "grad_norm": 0.11096387201363413,
      "learning_rate": 0.00012618997765657497,
      "loss": 1.0543,
      "step": 6010
    },
    {
      "epoch": 0.43328768110718663,
      "grad_norm": 0.13984362743377038,
      "learning_rate": 0.00012616744472239316,
      "loss": 1.0531,
      "step": 6011
    },
    {
      "epoch": 0.43335976356952355,
      "grad_norm": 0.24802510427819718,
      "learning_rate": 0.00012614491036185157,
      "loss": 1.1168,
      "step": 6012
    },
    {
      "epoch": 0.43343184603186047,
      "grad_norm": 0.12586597664326843,
      "learning_rate": 0.00012612237457617842,
      "loss": 1.0976,
      "step": 6013
    },
    {
      "epoch": 0.4335039284941974,
      "grad_norm": 0.11905569634904688,
      "learning_rate": 0.0001260998373666022,
      "loss": 1.012,
      "step": 6014
    },
    {
      "epoch": 0.43357601095653425,
      "grad_norm": 0.13242851153203686,
      "learning_rate": 0.0001260772987343513,
      "loss": 1.0033,
      "step": 6015
    },
    {
      "epoch": 0.43364809341887117,
      "grad_norm": 0.13280368645381743,
      "learning_rate": 0.00012605475868065436,
      "loss": 1.1295,
      "step": 6016
    },
    {
      "epoch": 0.4337201758812081,
      "grad_norm": 0.17406547715191403,
      "learning_rate": 0.00012603221720673997,
      "loss": 1.0365,
      "step": 6017
    },
    {
      "epoch": 0.433792258343545,
      "grad_norm": 0.48002990322685196,
      "learning_rate": 0.00012600967431383687,
      "loss": 1.1693,
      "step": 6018
    },
    {
      "epoch": 0.43386434080588193,
      "grad_norm": 0.15005020500036526,
      "learning_rate": 0.0001259871300031738,
      "loss": 1.129,
      "step": 6019
    },
    {
      "epoch": 0.43393642326821885,
      "grad_norm": 0.3299735436659352,
      "learning_rate": 0.00012596458427597968,
      "loss": 1.0904,
      "step": 6020
    },
    {
      "epoch": 0.43400850573055577,
      "grad_norm": 0.14768800776523414,
      "learning_rate": 0.00012594203713348342,
      "loss": 1.0895,
      "step": 6021
    },
    {
      "epoch": 0.4340805881928927,
      "grad_norm": 0.15777208192951142,
      "learning_rate": 0.00012591948857691406,
      "loss": 1.0605,
      "step": 6022
    },
    {
      "epoch": 0.43415267065522956,
      "grad_norm": 0.1661462682515543,
      "learning_rate": 0.00012589693860750066,
      "loss": 1.0937,
      "step": 6023
    },
    {
      "epoch": 0.4342247531175665,
      "grad_norm": 0.20143331563485833,
      "learning_rate": 0.00012587438722647248,
      "loss": 1.0309,
      "step": 6024
    },
    {
      "epoch": 0.4342968355799034,
      "grad_norm": 0.38195182199660144,
      "learning_rate": 0.00012585183443505869,
      "loss": 1.0871,
      "step": 6025
    },
    {
      "epoch": 0.4343689180422403,
      "grad_norm": 0.11859717093991665,
      "learning_rate": 0.00012582928023448863,
      "loss": 1.0398,
      "step": 6026
    },
    {
      "epoch": 0.43444100050457723,
      "grad_norm": 0.24218171209418424,
      "learning_rate": 0.00012580672462599173,
      "loss": 1.032,
      "step": 6027
    },
    {
      "epoch": 0.43451308296691415,
      "grad_norm": 0.12388813869706458,
      "learning_rate": 0.0001257841676107975,
      "loss": 1.0419,
      "step": 6028
    },
    {
      "epoch": 0.4345851654292511,
      "grad_norm": 0.11132182157553124,
      "learning_rate": 0.0001257616091901354,
      "loss": 1.0192,
      "step": 6029
    },
    {
      "epoch": 0.434657247891588,
      "grad_norm": 0.1584926932047678,
      "learning_rate": 0.00012573904936523516,
      "loss": 1.0055,
      "step": 6030
    },
    {
      "epoch": 0.4347293303539249,
      "grad_norm": 0.13949513060535954,
      "learning_rate": 0.00012571648813732646,
      "loss": 0.9695,
      "step": 6031
    },
    {
      "epoch": 0.4348014128162618,
      "grad_norm": 0.14069322058554135,
      "learning_rate": 0.00012569392550763913,
      "loss": 1.0567,
      "step": 6032
    },
    {
      "epoch": 0.4348734952785987,
      "grad_norm": 0.1280020486588746,
      "learning_rate": 0.00012567136147740294,
      "loss": 1.1187,
      "step": 6033
    },
    {
      "epoch": 0.4349455777409356,
      "grad_norm": 0.12331074583463057,
      "learning_rate": 0.0001256487960478479,
      "loss": 1.0562,
      "step": 6034
    },
    {
      "epoch": 0.43501766020327254,
      "grad_norm": 0.18323573482468353,
      "learning_rate": 0.000125626229220204,
      "loss": 1.1241,
      "step": 6035
    },
    {
      "epoch": 0.43508974266560946,
      "grad_norm": 0.11730835743710617,
      "learning_rate": 0.0001256036609957014,
      "loss": 1.0234,
      "step": 6036
    },
    {
      "epoch": 0.4351618251279464,
      "grad_norm": 0.1285299702832244,
      "learning_rate": 0.00012558109137557016,
      "loss": 1.0275,
      "step": 6037
    },
    {
      "epoch": 0.4352339075902833,
      "grad_norm": 0.11830956196007905,
      "learning_rate": 0.00012555852036104064,
      "loss": 1.0946,
      "step": 6038
    },
    {
      "epoch": 0.4353059900526202,
      "grad_norm": 0.11101947943820596,
      "learning_rate": 0.00012553594795334307,
      "loss": 1.0497,
      "step": 6039
    },
    {
      "epoch": 0.43537807251495714,
      "grad_norm": 0.13766321398064105,
      "learning_rate": 0.0001255133741537079,
      "loss": 0.9922,
      "step": 6040
    },
    {
      "epoch": 0.435450154977294,
      "grad_norm": 0.12818984096194402,
      "learning_rate": 0.00012549079896336557,
      "loss": 1.0467,
      "step": 6041
    },
    {
      "epoch": 0.4355222374396309,
      "grad_norm": 0.12269396349785343,
      "learning_rate": 0.00012546822238354666,
      "loss": 1.0027,
      "step": 6042
    },
    {
      "epoch": 0.43559431990196784,
      "grad_norm": 0.23217412888765152,
      "learning_rate": 0.00012544564441548182,
      "loss": 1.0884,
      "step": 6043
    },
    {
      "epoch": 0.43566640236430476,
      "grad_norm": 0.10411068149095018,
      "learning_rate": 0.00012542306506040167,
      "loss": 1.053,
      "step": 6044
    },
    {
      "epoch": 0.4357384848266417,
      "grad_norm": 0.3425221395153597,
      "learning_rate": 0.00012540048431953707,
      "loss": 1.0495,
      "step": 6045
    },
    {
      "epoch": 0.4358105672889786,
      "grad_norm": 0.9107760077367916,
      "learning_rate": 0.0001253779021941188,
      "loss": 0.9677,
      "step": 6046
    },
    {
      "epoch": 0.4358826497513155,
      "grad_norm": 0.10985299038224122,
      "learning_rate": 0.00012535531868537785,
      "loss": 1.0634,
      "step": 6047
    },
    {
      "epoch": 0.43595473221365244,
      "grad_norm": 0.14598467496015746,
      "learning_rate": 0.00012533273379454517,
      "loss": 0.999,
      "step": 6048
    },
    {
      "epoch": 0.4360268146759893,
      "grad_norm": 0.13662398067453466,
      "learning_rate": 0.00012531014752285188,
      "loss": 1.0241,
      "step": 6049
    },
    {
      "epoch": 0.4360988971383262,
      "grad_norm": 0.457049683831585,
      "learning_rate": 0.00012528755987152912,
      "loss": 1.0696,
      "step": 6050
    },
    {
      "epoch": 0.43617097960066314,
      "grad_norm": 0.24613763930042143,
      "learning_rate": 0.0001252649708418081,
      "loss": 1.0763,
      "step": 6051
    },
    {
      "epoch": 0.43624306206300006,
      "grad_norm": 0.13539566221599017,
      "learning_rate": 0.00012524238043492016,
      "loss": 1.1855,
      "step": 6052
    },
    {
      "epoch": 0.436315144525337,
      "grad_norm": 0.14272614209456694,
      "learning_rate": 0.00012521978865209668,
      "loss": 1.0211,
      "step": 6053
    },
    {
      "epoch": 0.4363872269876739,
      "grad_norm": 0.14066802130034747,
      "learning_rate": 0.0001251971954945691,
      "loss": 1.1297,
      "step": 6054
    },
    {
      "epoch": 0.4364593094500108,
      "grad_norm": 0.279561486986885,
      "learning_rate": 0.00012517460096356887,
      "loss": 1.1452,
      "step": 6055
    },
    {
      "epoch": 0.43653139191234774,
      "grad_norm": 0.1741713614901341,
      "learning_rate": 0.00012515200506032774,
      "loss": 1.0355,
      "step": 6056
    },
    {
      "epoch": 0.43660347437468466,
      "grad_norm": 0.1076851111831796,
      "learning_rate": 0.00012512940778607726,
      "loss": 1.0572,
      "step": 6057
    },
    {
      "epoch": 0.43667555683702153,
      "grad_norm": 0.16855299837712437,
      "learning_rate": 0.0001251068091420493,
      "loss": 1.0565,
      "step": 6058
    },
    {
      "epoch": 0.43674763929935845,
      "grad_norm": 0.12234341167221721,
      "learning_rate": 0.00012508420912947562,
      "loss": 1.0675,
      "step": 6059
    },
    {
      "epoch": 0.43681972176169537,
      "grad_norm": 0.11325066576338255,
      "learning_rate": 0.00012506160774958814,
      "loss": 0.9928,
      "step": 6060
    },
    {
      "epoch": 0.4368918042240323,
      "grad_norm": 0.11221578575298799,
      "learning_rate": 0.00012503900500361877,
      "loss": 1.0083,
      "step": 6061
    },
    {
      "epoch": 0.4369638866863692,
      "grad_norm": 0.11375898236854307,
      "learning_rate": 0.00012501640089279968,
      "loss": 1.0204,
      "step": 6062
    },
    {
      "epoch": 0.4370359691487061,
      "grad_norm": 0.13742586176029942,
      "learning_rate": 0.0001249937954183629,
      "loss": 1.0626,
      "step": 6063
    },
    {
      "epoch": 0.43710805161104305,
      "grad_norm": 0.14104937530733144,
      "learning_rate": 0.00012497118858154072,
      "loss": 1.0595,
      "step": 6064
    },
    {
      "epoch": 0.43718013407337997,
      "grad_norm": 0.13904262124599812,
      "learning_rate": 0.00012494858038356531,
      "loss": 1.083,
      "step": 6065
    },
    {
      "epoch": 0.4372522165357169,
      "grad_norm": 0.14591545668190753,
      "learning_rate": 0.0001249259708256691,
      "loss": 0.9813,
      "step": 6066
    },
    {
      "epoch": 0.43732429899805375,
      "grad_norm": 0.21193167658484335,
      "learning_rate": 0.00012490335990908452,
      "loss": 1.0482,
      "step": 6067
    },
    {
      "epoch": 0.43739638146039067,
      "grad_norm": 0.21364330168845494,
      "learning_rate": 0.00012488074763504397,
      "loss": 1.0772,
      "step": 6068
    },
    {
      "epoch": 0.4374684639227276,
      "grad_norm": 0.1521170941333164,
      "learning_rate": 0.0001248581340047801,
      "loss": 0.9903,
      "step": 6069
    },
    {
      "epoch": 0.4375405463850645,
      "grad_norm": 0.16546421798444136,
      "learning_rate": 0.00012483551901952558,
      "loss": 1.0104,
      "step": 6070
    },
    {
      "epoch": 0.43761262884740143,
      "grad_norm": 0.12215774927697161,
      "learning_rate": 0.00012481290268051304,
      "loss": 1.0375,
      "step": 6071
    },
    {
      "epoch": 0.43768471130973835,
      "grad_norm": 0.1637075477526935,
      "learning_rate": 0.00012479028498897535,
      "loss": 1.0134,
      "step": 6072
    },
    {
      "epoch": 0.43775679377207527,
      "grad_norm": 0.12762643746613994,
      "learning_rate": 0.00012476766594614534,
      "loss": 1.0962,
      "step": 6073
    },
    {
      "epoch": 0.4378288762344122,
      "grad_norm": 0.1648594185363139,
      "learning_rate": 0.00012474504555325594,
      "loss": 1.1271,
      "step": 6074
    },
    {
      "epoch": 0.43790095869674905,
      "grad_norm": 0.2890250593773366,
      "learning_rate": 0.0001247224238115402,
      "loss": 1.1127,
      "step": 6075
    },
    {
      "epoch": 0.437973041159086,
      "grad_norm": 0.1421418683451012,
      "learning_rate": 0.00012469980072223118,
      "loss": 1.0863,
      "step": 6076
    },
    {
      "epoch": 0.4380451236214229,
      "grad_norm": 0.1331827413465067,
      "learning_rate": 0.00012467717628656205,
      "loss": 1.0817,
      "step": 6077
    },
    {
      "epoch": 0.4381172060837598,
      "grad_norm": 0.10395769665583335,
      "learning_rate": 0.00012465455050576606,
      "loss": 1.043,
      "step": 6078
    },
    {
      "epoch": 0.43818928854609673,
      "grad_norm": 0.15874491439120772,
      "learning_rate": 0.00012463192338107648,
      "loss": 0.9941,
      "step": 6079
    },
    {
      "epoch": 0.43826137100843365,
      "grad_norm": 0.11296852700871311,
      "learning_rate": 0.00012460929491372674,
      "loss": 1.0758,
      "step": 6080
    },
    {
      "epoch": 0.4383334534707706,
      "grad_norm": 0.5355774018777789,
      "learning_rate": 0.00012458666510495022,
      "loss": 1.0804,
      "step": 6081
    },
    {
      "epoch": 0.4384055359331075,
      "grad_norm": 0.13038638143040573,
      "learning_rate": 0.00012456403395598052,
      "loss": 0.9903,
      "step": 6082
    },
    {
      "epoch": 0.4384776183954444,
      "grad_norm": 0.1402704679402766,
      "learning_rate": 0.00012454140146805125,
      "loss": 0.9752,
      "step": 6083
    },
    {
      "epoch": 0.4385497008577813,
      "grad_norm": 1.2079984261007775,
      "learning_rate": 0.000124518767642396,
      "loss": 1.0477,
      "step": 6084
    },
    {
      "epoch": 0.4386217833201182,
      "grad_norm": 0.11054226944969091,
      "learning_rate": 0.0001244961324802486,
      "loss": 1.0193,
      "step": 6085
    },
    {
      "epoch": 0.4386938657824551,
      "grad_norm": 0.156417115277778,
      "learning_rate": 0.00012447349598284277,
      "loss": 0.9984,
      "step": 6086
    },
    {
      "epoch": 0.43876594824479204,
      "grad_norm": 0.1423902176041196,
      "learning_rate": 0.00012445085815141252,
      "loss": 0.9824,
      "step": 6087
    },
    {
      "epoch": 0.43883803070712896,
      "grad_norm": 0.14506268033121827,
      "learning_rate": 0.00012442821898719174,
      "loss": 1.0739,
      "step": 6088
    },
    {
      "epoch": 0.4389101131694659,
      "grad_norm": 0.1223521464696027,
      "learning_rate": 0.00012440557849141448,
      "loss": 1.0719,
      "step": 6089
    },
    {
      "epoch": 0.4389821956318028,
      "grad_norm": 0.1317495862276151,
      "learning_rate": 0.00012438293666531484,
      "loss": 1.0805,
      "step": 6090
    },
    {
      "epoch": 0.4390542780941397,
      "grad_norm": 0.12376368290660186,
      "learning_rate": 0.00012436029351012705,
      "loss": 1.1203,
      "step": 6091
    },
    {
      "epoch": 0.43912636055647664,
      "grad_norm": 0.12590969848915703,
      "learning_rate": 0.0001243376490270853,
      "loss": 1.0728,
      "step": 6092
    },
    {
      "epoch": 0.4391984430188135,
      "grad_norm": 0.10566124128472797,
      "learning_rate": 0.00012431500321742395,
      "loss": 1.1211,
      "step": 6093
    },
    {
      "epoch": 0.4392705254811504,
      "grad_norm": 0.1670175335434736,
      "learning_rate": 0.0001242923560823774,
      "loss": 1.0799,
      "step": 6094
    },
    {
      "epoch": 0.43934260794348734,
      "grad_norm": 0.11799658947802373,
      "learning_rate": 0.00012426970762318014,
      "loss": 1.1032,
      "step": 6095
    },
    {
      "epoch": 0.43941469040582426,
      "grad_norm": 0.12217572381355278,
      "learning_rate": 0.00012424705784106664,
      "loss": 1.1052,
      "step": 6096
    },
    {
      "epoch": 0.4394867728681612,
      "grad_norm": 0.23736972108207258,
      "learning_rate": 0.00012422440673727164,
      "loss": 1.1175,
      "step": 6097
    },
    {
      "epoch": 0.4395588553304981,
      "grad_norm": 0.11003515791609998,
      "learning_rate": 0.0001242017543130297,
      "loss": 1.0612,
      "step": 6098
    },
    {
      "epoch": 0.439630937792835,
      "grad_norm": 0.1302334786144986,
      "learning_rate": 0.00012417910056957565,
      "loss": 1.0813,
      "step": 6099
    },
    {
      "epoch": 0.43970302025517194,
      "grad_norm": 0.11842522981985404,
      "learning_rate": 0.00012415644550814431,
      "loss": 1.0504,
      "step": 6100
    },
    {
      "epoch": 0.4397751027175088,
      "grad_norm": 0.18452107398065443,
      "learning_rate": 0.00012413378912997058,
      "loss": 0.9902,
      "step": 6101
    },
    {
      "epoch": 0.4398471851798457,
      "grad_norm": 0.10369598472052569,
      "learning_rate": 0.00012411113143628944,
      "loss": 1.0803,
      "step": 6102
    },
    {
      "epoch": 0.43991926764218264,
      "grad_norm": 0.8244488992203503,
      "learning_rate": 0.0001240884724283359,
      "loss": 1.0453,
      "step": 6103
    },
    {
      "epoch": 0.43999135010451956,
      "grad_norm": 0.12331332164651046,
      "learning_rate": 0.00012406581210734513,
      "loss": 1.0028,
      "step": 6104
    },
    {
      "epoch": 0.4400634325668565,
      "grad_norm": 0.1425793791757239,
      "learning_rate": 0.00012404315047455234,
      "loss": 1.0659,
      "step": 6105
    },
    {
      "epoch": 0.4401355150291934,
      "grad_norm": 0.11435723891648421,
      "learning_rate": 0.0001240204875311927,
      "loss": 1.0603,
      "step": 6106
    },
    {
      "epoch": 0.4402075974915303,
      "grad_norm": 0.12714169313303866,
      "learning_rate": 0.00012399782327850162,
      "loss": 0.9977,
      "step": 6107
    },
    {
      "epoch": 0.44027967995386724,
      "grad_norm": 0.12020017232161345,
      "learning_rate": 0.00012397515771771448,
      "loss": 1.068,
      "step": 6108
    },
    {
      "epoch": 0.44035176241620416,
      "grad_norm": 0.12211556057715382,
      "learning_rate": 0.00012395249085006676,
      "loss": 1.0276,
      "step": 6109
    },
    {
      "epoch": 0.440423844878541,
      "grad_norm": 0.12300563782368963,
      "learning_rate": 0.00012392982267679398,
      "loss": 1.004,
      "step": 6110
    },
    {
      "epoch": 0.44049592734087795,
      "grad_norm": 0.20811989428971894,
      "learning_rate": 0.00012390715319913182,
      "loss": 1.1118,
      "step": 6111
    },
    {
      "epoch": 0.44056800980321487,
      "grad_norm": 0.14090900893126787,
      "learning_rate": 0.0001238844824183159,
      "loss": 1.1101,
      "step": 6112
    },
    {
      "epoch": 0.4406400922655518,
      "grad_norm": 0.3059610028409089,
      "learning_rate": 0.00012386181033558204,
      "loss": 1.0024,
      "step": 6113
    },
    {
      "epoch": 0.4407121747278887,
      "grad_norm": 0.16882986356515253,
      "learning_rate": 0.000123839136952166,
      "loss": 1.0313,
      "step": 6114
    },
    {
      "epoch": 0.4407842571902256,
      "grad_norm": 0.1397168749875687,
      "learning_rate": 0.00012381646226930376,
      "loss": 1.0355,
      "step": 6115
    },
    {
      "epoch": 0.44085633965256255,
      "grad_norm": 0.14556467922163385,
      "learning_rate": 0.00012379378628823125,
      "loss": 1.0336,
      "step": 6116
    },
    {
      "epoch": 0.44092842211489947,
      "grad_norm": 0.11838508052279649,
      "learning_rate": 0.0001237711090101845,
      "loss": 1.0631,
      "step": 6117
    },
    {
      "epoch": 0.4410005045772364,
      "grad_norm": 0.12786173127167255,
      "learning_rate": 0.00012374843043639965,
      "loss": 1.123,
      "step": 6118
    },
    {
      "epoch": 0.44107258703957325,
      "grad_norm": 0.1120750487800199,
      "learning_rate": 0.00012372575056811294,
      "loss": 1.0666,
      "step": 6119
    },
    {
      "epoch": 0.44114466950191017,
      "grad_norm": 0.13540285057614707,
      "learning_rate": 0.00012370306940656051,
      "loss": 1.0486,
      "step": 6120
    },
    {
      "epoch": 0.4412167519642471,
      "grad_norm": 0.11991092885559941,
      "learning_rate": 0.00012368038695297876,
      "loss": 1.0302,
      "step": 6121
    },
    {
      "epoch": 0.441288834426584,
      "grad_norm": 0.1593863875405006,
      "learning_rate": 0.00012365770320860406,
      "loss": 1.0502,
      "step": 6122
    },
    {
      "epoch": 0.44136091688892093,
      "grad_norm": 0.12559711878054117,
      "learning_rate": 0.0001236350181746729,
      "loss": 1.0462,
      "step": 6123
    },
    {
      "epoch": 0.44143299935125785,
      "grad_norm": 0.30568109066184007,
      "learning_rate": 0.0001236123318524218,
      "loss": 1.1114,
      "step": 6124
    },
    {
      "epoch": 0.44150508181359477,
      "grad_norm": 0.14838028493462865,
      "learning_rate": 0.00012358964424308738,
      "loss": 1.0646,
      "step": 6125
    },
    {
      "epoch": 0.4415771642759317,
      "grad_norm": 0.10899706767562214,
      "learning_rate": 0.0001235669553479063,
      "loss": 1.0408,
      "step": 6126
    },
    {
      "epoch": 0.44164924673826855,
      "grad_norm": 0.12398017977945347,
      "learning_rate": 0.00012354426516811534,
      "loss": 1.0599,
      "step": 6127
    },
    {
      "epoch": 0.4417213292006055,
      "grad_norm": 0.11090073461956983,
      "learning_rate": 0.00012352157370495125,
      "loss": 1.0166,
      "step": 6128
    },
    {
      "epoch": 0.4417934116629424,
      "grad_norm": 0.13679685951880435,
      "learning_rate": 0.00012349888095965097,
      "loss": 0.9716,
      "step": 6129
    },
    {
      "epoch": 0.4418654941252793,
      "grad_norm": 0.12292960287980503,
      "learning_rate": 0.0001234761869334515,
      "loss": 1.066,
      "step": 6130
    },
    {
      "epoch": 0.44193757658761623,
      "grad_norm": 0.1355482370230008,
      "learning_rate": 0.00012345349162758978,
      "loss": 1.0149,
      "step": 6131
    },
    {
      "epoch": 0.44200965904995315,
      "grad_norm": 0.12018971521574304,
      "learning_rate": 0.00012343079504330295,
      "loss": 1.0951,
      "step": 6132
    },
    {
      "epoch": 0.44208174151229007,
      "grad_norm": 0.18536789028304132,
      "learning_rate": 0.00012340809718182819,
      "loss": 1.0285,
      "step": 6133
    },
    {
      "epoch": 0.442153823974627,
      "grad_norm": 0.16050198602179855,
      "learning_rate": 0.00012338539804440267,
      "loss": 1.0603,
      "step": 6134
    },
    {
      "epoch": 0.4422259064369639,
      "grad_norm": 0.11288562690365163,
      "learning_rate": 0.00012336269763226382,
      "loss": 1.0338,
      "step": 6135
    },
    {
      "epoch": 0.4422979888993008,
      "grad_norm": 0.11376362137848073,
      "learning_rate": 0.00012333999594664886,
      "loss": 1.0393,
      "step": 6136
    },
    {
      "epoch": 0.4423700713616377,
      "grad_norm": 0.12169439217758514,
      "learning_rate": 0.0001233172929887954,
      "loss": 1.0861,
      "step": 6137
    },
    {
      "epoch": 0.4424421538239746,
      "grad_norm": 0.11841153149183349,
      "learning_rate": 0.0001232945887599408,
      "loss": 1.0655,
      "step": 6138
    },
    {
      "epoch": 0.44251423628631154,
      "grad_norm": 0.12439445412908566,
      "learning_rate": 0.00012327188326132274,
      "loss": 1.0573,
      "step": 6139
    },
    {
      "epoch": 0.44258631874864846,
      "grad_norm": 0.14466649638959259,
      "learning_rate": 0.0001232491764941788,
      "loss": 1.0257,
      "step": 6140
    },
    {
      "epoch": 0.4426584012109854,
      "grad_norm": 0.12268014421070597,
      "learning_rate": 0.00012322646845974684,
      "loss": 1.0718,
      "step": 6141
    },
    {
      "epoch": 0.4427304836733223,
      "grad_norm": 0.11853224103095274,
      "learning_rate": 0.0001232037591592645,
      "loss": 1.0294,
      "step": 6142
    },
    {
      "epoch": 0.4428025661356592,
      "grad_norm": 0.30751741934475507,
      "learning_rate": 0.00012318104859396973,
      "loss": 1.0405,
      "step": 6143
    },
    {
      "epoch": 0.44287464859799613,
      "grad_norm": 0.1411224757444492,
      "learning_rate": 0.0001231583367651004,
      "loss": 1.0451,
      "step": 6144
    },
    {
      "epoch": 0.442946731060333,
      "grad_norm": 1.4651738827947964,
      "learning_rate": 0.00012313562367389453,
      "loss": 1.0738,
      "step": 6145
    },
    {
      "epoch": 0.4430188135226699,
      "grad_norm": 0.12172398951261189,
      "learning_rate": 0.0001231129093215902,
      "loss": 0.9872,
      "step": 6146
    },
    {
      "epoch": 0.44309089598500684,
      "grad_norm": 0.5590398676384626,
      "learning_rate": 0.00012309019370942555,
      "loss": 1.0334,
      "step": 6147
    },
    {
      "epoch": 0.44316297844734376,
      "grad_norm": 0.29406553758783077,
      "learning_rate": 0.00012306747683863878,
      "loss": 1.0973,
      "step": 6148
    },
    {
      "epoch": 0.4432350609096807,
      "grad_norm": 0.21342175335080413,
      "learning_rate": 0.00012304475871046812,
      "loss": 1.1358,
      "step": 6149
    },
    {
      "epoch": 0.4433071433720176,
      "grad_norm": 0.12675872268064514,
      "learning_rate": 0.00012302203932615198,
      "loss": 1.0628,
      "step": 6150
    },
    {
      "epoch": 0.4433792258343545,
      "grad_norm": 0.10723648440100515,
      "learning_rate": 0.0001229993186869287,
      "loss": 0.9832,
      "step": 6151
    },
    {
      "epoch": 0.44345130829669144,
      "grad_norm": 0.12065363386031652,
      "learning_rate": 0.00012297659679403683,
      "loss": 1.1098,
      "step": 6152
    },
    {
      "epoch": 0.4435233907590283,
      "grad_norm": 0.25757392266557183,
      "learning_rate": 0.00012295387364871487,
      "loss": 1.108,
      "step": 6153
    },
    {
      "epoch": 0.4435954732213652,
      "grad_norm": 0.1363830943185448,
      "learning_rate": 0.00012293114925220145,
      "loss": 1.0856,
      "step": 6154
    },
    {
      "epoch": 0.44366755568370214,
      "grad_norm": 0.19670542405964048,
      "learning_rate": 0.00012290842360573523,
      "loss": 1.0064,
      "step": 6155
    },
    {
      "epoch": 0.44373963814603906,
      "grad_norm": 0.135290616282325,
      "learning_rate": 0.00012288569671055498,
      "loss": 1.0185,
      "step": 6156
    },
    {
      "epoch": 0.443811720608376,
      "grad_norm": 0.14633886998958154,
      "learning_rate": 0.00012286296856789953,
      "loss": 1.0751,
      "step": 6157
    },
    {
      "epoch": 0.4438838030707129,
      "grad_norm": 0.6898875761779532,
      "learning_rate": 0.00012284023917900775,
      "loss": 1.07,
      "step": 6158
    },
    {
      "epoch": 0.4439558855330498,
      "grad_norm": 0.1168511662492843,
      "learning_rate": 0.0001228175085451186,
      "loss": 1.0368,
      "step": 6159
    },
    {
      "epoch": 0.44402796799538674,
      "grad_norm": 0.1146873443465321,
      "learning_rate": 0.0001227947766674711,
      "loss": 1.0743,
      "step": 6160
    },
    {
      "epoch": 0.44410005045772366,
      "grad_norm": 0.7024715152215398,
      "learning_rate": 0.00012277204354730435,
      "loss": 1.0949,
      "step": 6161
    },
    {
      "epoch": 0.4441721329200605,
      "grad_norm": 0.16517028151299062,
      "learning_rate": 0.00012274930918585749,
      "loss": 1.0642,
      "step": 6162
    },
    {
      "epoch": 0.44424421538239744,
      "grad_norm": 1.3877034853434094,
      "learning_rate": 0.00012272657358436978,
      "loss": 1.0613,
      "step": 6163
    },
    {
      "epoch": 0.44431629784473436,
      "grad_norm": 0.18114958917746812,
      "learning_rate": 0.0001227038367440805,
      "loss": 1.0595,
      "step": 6164
    },
    {
      "epoch": 0.4443883803070713,
      "grad_norm": 0.9317472136853716,
      "learning_rate": 0.00012268109866622897,
      "loss": 1.0278,
      "step": 6165
    },
    {
      "epoch": 0.4444604627694082,
      "grad_norm": 0.3548979249199682,
      "learning_rate": 0.00012265835935205467,
      "loss": 1.0323,
      "step": 6166
    },
    {
      "epoch": 0.4445325452317451,
      "grad_norm": 0.23226999553035058,
      "learning_rate": 0.00012263561880279707,
      "loss": 0.9749,
      "step": 6167
    },
    {
      "epoch": 0.44460462769408204,
      "grad_norm": 0.16359608673949602,
      "learning_rate": 0.00012261287701969574,
      "loss": 0.9953,
      "step": 6168
    },
    {
      "epoch": 0.44467671015641896,
      "grad_norm": 0.1398328610316583,
      "learning_rate": 0.0001225901340039903,
      "loss": 1.0945,
      "step": 6169
    },
    {
      "epoch": 0.4447487926187559,
      "grad_norm": 0.15734420245676187,
      "learning_rate": 0.0001225673897569205,
      "loss": 1.0897,
      "step": 6170
    },
    {
      "epoch": 0.44482087508109275,
      "grad_norm": 0.16531386146900956,
      "learning_rate": 0.00012254464427972602,
      "loss": 1.0316,
      "step": 6171
    },
    {
      "epoch": 0.44489295754342967,
      "grad_norm": 0.2112817554677162,
      "learning_rate": 0.00012252189757364678,
      "loss": 1.0311,
      "step": 6172
    },
    {
      "epoch": 0.4449650400057666,
      "grad_norm": 0.17110779635012577,
      "learning_rate": 0.0001224991496399226,
      "loss": 1.1113,
      "step": 6173
    },
    {
      "epoch": 0.4450371224681035,
      "grad_norm": 0.1752399598352518,
      "learning_rate": 0.0001224764004797935,
      "loss": 1.0865,
      "step": 6174
    },
    {
      "epoch": 0.4451092049304404,
      "grad_norm": 0.19182902761717693,
      "learning_rate": 0.0001224536500944995,
      "loss": 1.103,
      "step": 6175
    },
    {
      "epoch": 0.44518128739277735,
      "grad_norm": 0.18314358727363453,
      "learning_rate": 0.00012243089848528067,
      "loss": 1.0306,
      "step": 6176
    },
    {
      "epoch": 0.44525336985511427,
      "grad_norm": 0.13548004187142396,
      "learning_rate": 0.00012240814565337723,
      "loss": 1.0182,
      "step": 6177
    },
    {
      "epoch": 0.4453254523174512,
      "grad_norm": 0.12979563207722453,
      "learning_rate": 0.00012238539160002934,
      "loss": 1.0649,
      "step": 6178
    },
    {
      "epoch": 0.44539753477978805,
      "grad_norm": 0.13841316848353696,
      "learning_rate": 0.00012236263632647737,
      "loss": 1.0542,
      "step": 6179
    },
    {
      "epoch": 0.44546961724212497,
      "grad_norm": 0.1457517314260125,
      "learning_rate": 0.00012233987983396166,
      "loss": 1.101,
      "step": 6180
    },
    {
      "epoch": 0.4455416997044619,
      "grad_norm": 0.22599468606890025,
      "learning_rate": 0.00012231712212372263,
      "loss": 1.1126,
      "step": 6181
    },
    {
      "epoch": 0.4456137821667988,
      "grad_norm": 0.1475178582822439,
      "learning_rate": 0.0001222943631970008,
      "loss": 1.0583,
      "step": 6182
    },
    {
      "epoch": 0.44568586462913573,
      "grad_norm": 0.23458883733421299,
      "learning_rate": 0.0001222716030550367,
      "loss": 1.0471,
      "step": 6183
    },
    {
      "epoch": 0.44575794709147265,
      "grad_norm": 0.3214584449441944,
      "learning_rate": 0.00012224884169907105,
      "loss": 1.0755,
      "step": 6184
    },
    {
      "epoch": 0.44583002955380957,
      "grad_norm": 0.1330649291456206,
      "learning_rate": 0.0001222260791303444,
      "loss": 0.9984,
      "step": 6185
    },
    {
      "epoch": 0.4459021120161465,
      "grad_norm": 0.13063088348039995,
      "learning_rate": 0.0001222033153500977,
      "loss": 1.1097,
      "step": 6186
    },
    {
      "epoch": 0.4459741944784834,
      "grad_norm": 0.11893514449630667,
      "learning_rate": 0.00012218055035957155,
      "loss": 1.0848,
      "step": 6187
    },
    {
      "epoch": 0.4460462769408203,
      "grad_norm": 0.21460833323219505,
      "learning_rate": 0.00012215778416000707,
      "loss": 1.0096,
      "step": 6188
    },
    {
      "epoch": 0.4461183594031572,
      "grad_norm": 0.20990235442504998,
      "learning_rate": 0.00012213501675264508,
      "loss": 1.1299,
      "step": 6189
    },
    {
      "epoch": 0.4461904418654941,
      "grad_norm": 0.12237823908162714,
      "learning_rate": 0.00012211224813872667,
      "loss": 1.059,
      "step": 6190
    },
    {
      "epoch": 0.44626252432783103,
      "grad_norm": 0.1332693637812345,
      "learning_rate": 0.00012208947831949288,
      "loss": 1.0174,
      "step": 6191
    },
    {
      "epoch": 0.44633460679016795,
      "grad_norm": 0.38874570221728316,
      "learning_rate": 0.00012206670729618497,
      "loss": 1.1213,
      "step": 6192
    },
    {
      "epoch": 0.4464066892525049,
      "grad_norm": 0.15433250108157265,
      "learning_rate": 0.00012204393507004403,
      "loss": 1.0213,
      "step": 6193
    },
    {
      "epoch": 0.4464787717148418,
      "grad_norm": 0.13748268723890925,
      "learning_rate": 0.00012202116164231145,
      "loss": 1.0573,
      "step": 6194
    },
    {
      "epoch": 0.4465508541771787,
      "grad_norm": 0.15438952861750943,
      "learning_rate": 0.00012199838701422856,
      "loss": 1.0701,
      "step": 6195
    },
    {
      "epoch": 0.44662293663951563,
      "grad_norm": 0.11834281012181196,
      "learning_rate": 0.00012197561118703678,
      "loss": 1.0465,
      "step": 6196
    },
    {
      "epoch": 0.4466950191018525,
      "grad_norm": 0.12935935608423124,
      "learning_rate": 0.00012195283416197756,
      "loss": 1.0863,
      "step": 6197
    },
    {
      "epoch": 0.4467671015641894,
      "grad_norm": 0.14387844579235456,
      "learning_rate": 0.0001219300559402925,
      "loss": 1.0029,
      "step": 6198
    },
    {
      "epoch": 0.44683918402652634,
      "grad_norm": 0.19568890715576887,
      "learning_rate": 0.00012190727652322321,
      "loss": 1.0098,
      "step": 6199
    },
    {
      "epoch": 0.44691126648886326,
      "grad_norm": 0.116181525072673,
      "learning_rate": 0.00012188449591201134,
      "loss": 1.1138,
      "step": 6200
    },
    {
      "epoch": 0.4469833489512002,
      "grad_norm": 0.1383251473002152,
      "learning_rate": 0.00012186171410789866,
      "loss": 1.0685,
      "step": 6201
    },
    {
      "epoch": 0.4470554314135371,
      "grad_norm": 0.4044956189662557,
      "learning_rate": 0.00012183893111212698,
      "loss": 1.1088,
      "step": 6202
    },
    {
      "epoch": 0.447127513875874,
      "grad_norm": 0.1118055489425271,
      "learning_rate": 0.00012181614692593819,
      "loss": 1.0213,
      "step": 6203
    },
    {
      "epoch": 0.44719959633821094,
      "grad_norm": 0.12248187912511589,
      "learning_rate": 0.00012179336155057422,
      "loss": 1.1425,
      "step": 6204
    },
    {
      "epoch": 0.4472716788005478,
      "grad_norm": 0.24745959934426548,
      "learning_rate": 0.00012177057498727706,
      "loss": 1.0477,
      "step": 6205
    },
    {
      "epoch": 0.4473437612628847,
      "grad_norm": 0.11953650159023552,
      "learning_rate": 0.00012174778723728884,
      "loss": 1.1235,
      "step": 6206
    },
    {
      "epoch": 0.44741584372522164,
      "grad_norm": 0.27091805032522837,
      "learning_rate": 0.0001217249983018516,
      "loss": 1.0781,
      "step": 6207
    },
    {
      "epoch": 0.44748792618755856,
      "grad_norm": 0.16551437193318208,
      "learning_rate": 0.00012170220818220763,
      "loss": 1.0657,
      "step": 6208
    },
    {
      "epoch": 0.4475600086498955,
      "grad_norm": 0.17901894523378187,
      "learning_rate": 0.00012167941687959914,
      "loss": 1.0779,
      "step": 6209
    },
    {
      "epoch": 0.4476320911122324,
      "grad_norm": 0.13567665876014923,
      "learning_rate": 0.00012165662439526851,
      "loss": 1.0421,
      "step": 6210
    },
    {
      "epoch": 0.4477041735745693,
      "grad_norm": 0.1269467411579209,
      "learning_rate": 0.00012163383073045806,
      "loss": 1.0094,
      "step": 6211
    },
    {
      "epoch": 0.44777625603690624,
      "grad_norm": 0.1721538030267381,
      "learning_rate": 0.00012161103588641034,
      "loss": 1.0839,
      "step": 6212
    },
    {
      "epoch": 0.44784833849924316,
      "grad_norm": 0.20201391039143737,
      "learning_rate": 0.00012158823986436778,
      "loss": 1.0601,
      "step": 6213
    },
    {
      "epoch": 0.44792042096158,
      "grad_norm": 0.2550882837377736,
      "learning_rate": 0.00012156544266557305,
      "loss": 1.0556,
      "step": 6214
    },
    {
      "epoch": 0.44799250342391694,
      "grad_norm": 0.14320736386314098,
      "learning_rate": 0.00012154264429126874,
      "loss": 1.0309,
      "step": 6215
    },
    {
      "epoch": 0.44806458588625386,
      "grad_norm": 0.13031796622980196,
      "learning_rate": 0.00012151984474269761,
      "loss": 1.0907,
      "step": 6216
    },
    {
      "epoch": 0.4481366683485908,
      "grad_norm": 0.11644312078412167,
      "learning_rate": 0.00012149704402110243,
      "loss": 1.0072,
      "step": 6217
    },
    {
      "epoch": 0.4482087508109277,
      "grad_norm": 0.4087539789185473,
      "learning_rate": 0.000121474242127726,
      "loss": 1.0523,
      "step": 6218
    },
    {
      "epoch": 0.4482808332732646,
      "grad_norm": 0.14012669122457452,
      "learning_rate": 0.0001214514390638113,
      "loss": 1.0303,
      "step": 6219
    },
    {
      "epoch": 0.44835291573560154,
      "grad_norm": 0.13796675562452718,
      "learning_rate": 0.00012142863483060123,
      "loss": 0.973,
      "step": 6220
    },
    {
      "epoch": 0.44842499819793846,
      "grad_norm": 0.17516091848726778,
      "learning_rate": 0.00012140582942933884,
      "loss": 1.14,
      "step": 6221
    },
    {
      "epoch": 0.4484970806602754,
      "grad_norm": 0.1322884069631158,
      "learning_rate": 0.00012138302286126728,
      "loss": 1.0586,
      "step": 6222
    },
    {
      "epoch": 0.44856916312261225,
      "grad_norm": 0.1558964194950442,
      "learning_rate": 0.00012136021512762965,
      "loss": 1.0704,
      "step": 6223
    },
    {
      "epoch": 0.44864124558494917,
      "grad_norm": 0.1539733058238346,
      "learning_rate": 0.00012133740622966923,
      "loss": 0.988,
      "step": 6224
    },
    {
      "epoch": 0.4487133280472861,
      "grad_norm": 0.1638272123099864,
      "learning_rate": 0.00012131459616862926,
      "loss": 0.9962,
      "step": 6225
    },
    {
      "epoch": 0.448785410509623,
      "grad_norm": 0.272722090934583,
      "learning_rate": 0.0001212917849457531,
      "loss": 1.0728,
      "step": 6226
    },
    {
      "epoch": 0.4488574929719599,
      "grad_norm": 2.4143003814862367,
      "learning_rate": 0.0001212689725622842,
      "loss": 1.0956,
      "step": 6227
    },
    {
      "epoch": 0.44892957543429685,
      "grad_norm": 0.1823943309493837,
      "learning_rate": 0.00012124615901946601,
      "loss": 1.0131,
      "step": 6228
    },
    {
      "epoch": 0.44900165789663377,
      "grad_norm": 0.1254936715877871,
      "learning_rate": 0.0001212233443185421,
      "loss": 1.0309,
      "step": 6229
    },
    {
      "epoch": 0.4490737403589707,
      "grad_norm": 0.13146552599403477,
      "learning_rate": 0.00012120052846075602,
      "loss": 0.9594,
      "step": 6230
    },
    {
      "epoch": 0.44914582282130755,
      "grad_norm": 0.22055841151378636,
      "learning_rate": 0.0001211777114473515,
      "loss": 1.0286,
      "step": 6231
    },
    {
      "epoch": 0.44921790528364447,
      "grad_norm": 0.12496326712180633,
      "learning_rate": 0.00012115489327957223,
      "loss": 1.071,
      "step": 6232
    },
    {
      "epoch": 0.4492899877459814,
      "grad_norm": 0.15258940788627637,
      "learning_rate": 0.00012113207395866204,
      "loss": 1.0624,
      "step": 6233
    },
    {
      "epoch": 0.4493620702083183,
      "grad_norm": 0.1276063919142963,
      "learning_rate": 0.00012110925348586475,
      "loss": 0.9968,
      "step": 6234
    },
    {
      "epoch": 0.44943415267065523,
      "grad_norm": 0.12940419279836088,
      "learning_rate": 0.00012108643186242429,
      "loss": 0.9925,
      "step": 6235
    },
    {
      "epoch": 0.44950623513299215,
      "grad_norm": 0.3829366315154146,
      "learning_rate": 0.00012106360908958465,
      "loss": 1.0401,
      "step": 6236
    },
    {
      "epoch": 0.44957831759532907,
      "grad_norm": 0.11723833601893213,
      "learning_rate": 0.0001210407851685899,
      "loss": 1.0568,
      "step": 6237
    },
    {
      "epoch": 0.449650400057666,
      "grad_norm": 0.15033846170364557,
      "learning_rate": 0.00012101796010068409,
      "loss": 1.1068,
      "step": 6238
    },
    {
      "epoch": 0.4497224825200029,
      "grad_norm": 0.13267089620695816,
      "learning_rate": 0.00012099513388711146,
      "loss": 1.0699,
      "step": 6239
    },
    {
      "epoch": 0.4497945649823398,
      "grad_norm": 0.17958607087470568,
      "learning_rate": 0.00012097230652911617,
      "loss": 1.134,
      "step": 6240
    },
    {
      "epoch": 0.4498666474446767,
      "grad_norm": 0.2993237908513895,
      "learning_rate": 0.00012094947802794258,
      "loss": 1.0528,
      "step": 6241
    },
    {
      "epoch": 0.4499387299070136,
      "grad_norm": 0.25841229758857714,
      "learning_rate": 0.00012092664838483499,
      "loss": 1.0017,
      "step": 6242
    },
    {
      "epoch": 0.45001081236935053,
      "grad_norm": 0.12998613364367206,
      "learning_rate": 0.00012090381760103786,
      "loss": 1.0375,
      "step": 6243
    },
    {
      "epoch": 0.45008289483168745,
      "grad_norm": 0.13880815531901924,
      "learning_rate": 0.00012088098567779564,
      "loss": 1.0886,
      "step": 6244
    },
    {
      "epoch": 0.45015497729402437,
      "grad_norm": 0.15428095698173863,
      "learning_rate": 0.00012085815261635293,
      "loss": 1.0732,
      "step": 6245
    },
    {
      "epoch": 0.4502270597563613,
      "grad_norm": 0.15098109566840065,
      "learning_rate": 0.00012083531841795425,
      "loss": 1.0371,
      "step": 6246
    },
    {
      "epoch": 0.4502991422186982,
      "grad_norm": 0.12116532521535583,
      "learning_rate": 0.00012081248308384438,
      "loss": 1.0317,
      "step": 6247
    },
    {
      "epoch": 0.45037122468103513,
      "grad_norm": 0.13082536881954973,
      "learning_rate": 0.00012078964661526795,
      "loss": 1.0412,
      "step": 6248
    },
    {
      "epoch": 0.450443307143372,
      "grad_norm": 0.28061511872402417,
      "learning_rate": 0.00012076680901346981,
      "loss": 1.0872,
      "step": 6249
    },
    {
      "epoch": 0.4505153896057089,
      "grad_norm": 0.1724557916615735,
      "learning_rate": 0.00012074397027969477,
      "loss": 1.0951,
      "step": 6250
    },
    {
      "epoch": 0.45058747206804584,
      "grad_norm": 0.16305771417271125,
      "learning_rate": 0.0001207211304151878,
      "loss": 1.0371,
      "step": 6251
    },
    {
      "epoch": 0.45065955453038276,
      "grad_norm": 0.12877012352838962,
      "learning_rate": 0.00012069828942119381,
      "loss": 1.1178,
      "step": 6252
    },
    {
      "epoch": 0.4507316369927197,
      "grad_norm": 0.15670174892572802,
      "learning_rate": 0.0001206754472989579,
      "loss": 1.0993,
      "step": 6253
    },
    {
      "epoch": 0.4508037194550566,
      "grad_norm": 0.144037693371522,
      "learning_rate": 0.00012065260404972514,
      "loss": 1.0248,
      "step": 6254
    },
    {
      "epoch": 0.4508758019173935,
      "grad_norm": 0.1234731960115264,
      "learning_rate": 0.00012062975967474069,
      "loss": 1.075,
      "step": 6255
    },
    {
      "epoch": 0.45094788437973043,
      "grad_norm": 0.16783207543261205,
      "learning_rate": 0.00012060691417524976,
      "loss": 1.0322,
      "step": 6256
    },
    {
      "epoch": 0.4510199668420673,
      "grad_norm": 0.12027707243913188,
      "learning_rate": 0.00012058406755249769,
      "loss": 1.001,
      "step": 6257
    },
    {
      "epoch": 0.4510920493044042,
      "grad_norm": 0.12978459162852574,
      "learning_rate": 0.00012056121980772977,
      "loss": 1.05,
      "step": 6258
    },
    {
      "epoch": 0.45116413176674114,
      "grad_norm": 0.1442908929835397,
      "learning_rate": 0.00012053837094219142,
      "loss": 1.1441,
      "step": 6259
    },
    {
      "epoch": 0.45123621422907806,
      "grad_norm": 0.13389377004572325,
      "learning_rate": 0.00012051552095712812,
      "loss": 1.112,
      "step": 6260
    },
    {
      "epoch": 0.451308296691415,
      "grad_norm": 0.1341802797072918,
      "learning_rate": 0.00012049266985378541,
      "loss": 1.0327,
      "step": 6261
    },
    {
      "epoch": 0.4513803791537519,
      "grad_norm": 0.12257536142836724,
      "learning_rate": 0.00012046981763340882,
      "loss": 1.0406,
      "step": 6262
    },
    {
      "epoch": 0.4514524616160888,
      "grad_norm": 0.14422333700220547,
      "learning_rate": 0.00012044696429724407,
      "loss": 1.0436,
      "step": 6263
    },
    {
      "epoch": 0.45152454407842574,
      "grad_norm": 0.20475964765238197,
      "learning_rate": 0.0001204241098465368,
      "loss": 0.9973,
      "step": 6264
    },
    {
      "epoch": 0.45159662654076266,
      "grad_norm": 0.11109868301536442,
      "learning_rate": 0.00012040125428253288,
      "loss": 1.0138,
      "step": 6265
    },
    {
      "epoch": 0.4516687090030995,
      "grad_norm": 0.14292996764606317,
      "learning_rate": 0.00012037839760647805,
      "loss": 1.093,
      "step": 6266
    },
    {
      "epoch": 0.45174079146543644,
      "grad_norm": 0.18416056186101112,
      "learning_rate": 0.00012035553981961827,
      "loss": 1.1374,
      "step": 6267
    },
    {
      "epoch": 0.45181287392777336,
      "grad_norm": 0.13011237481665225,
      "learning_rate": 0.00012033268092319943,
      "loss": 1.0012,
      "step": 6268
    },
    {
      "epoch": 0.4518849563901103,
      "grad_norm": 0.1290065375472841,
      "learning_rate": 0.0001203098209184676,
      "loss": 0.996,
      "step": 6269
    },
    {
      "epoch": 0.4519570388524472,
      "grad_norm": 0.5299158926907845,
      "learning_rate": 0.00012028695980666882,
      "loss": 1.0395,
      "step": 6270
    },
    {
      "epoch": 0.4520291213147841,
      "grad_norm": 0.13644118233191763,
      "learning_rate": 0.00012026409758904927,
      "loss": 1.1157,
      "step": 6271
    },
    {
      "epoch": 0.45210120377712104,
      "grad_norm": 0.12918928294402915,
      "learning_rate": 0.00012024123426685508,
      "loss": 1.1107,
      "step": 6272
    },
    {
      "epoch": 0.45217328623945796,
      "grad_norm": 0.16412807670320068,
      "learning_rate": 0.00012021836984133255,
      "loss": 1.0811,
      "step": 6273
    },
    {
      "epoch": 0.4522453687017949,
      "grad_norm": 0.16183428190418167,
      "learning_rate": 0.000120195504313728,
      "loss": 1.0626,
      "step": 6274
    },
    {
      "epoch": 0.45231745116413175,
      "grad_norm": 0.14673682214951217,
      "learning_rate": 0.00012017263768528775,
      "loss": 1.0508,
      "step": 6275
    },
    {
      "epoch": 0.45238953362646867,
      "grad_norm": 0.15025487663399617,
      "learning_rate": 0.0001201497699572583,
      "loss": 1.0046,
      "step": 6276
    },
    {
      "epoch": 0.4524616160888056,
      "grad_norm": 0.1511652204225981,
      "learning_rate": 0.00012012690113088613,
      "loss": 1.1066,
      "step": 6277
    },
    {
      "epoch": 0.4525336985511425,
      "grad_norm": 0.27294419368713746,
      "learning_rate": 0.00012010403120741778,
      "loss": 1.0144,
      "step": 6278
    },
    {
      "epoch": 0.4526057810134794,
      "grad_norm": 0.16654131267705563,
      "learning_rate": 0.00012008116018809988,
      "loss": 1.0137,
      "step": 6279
    },
    {
      "epoch": 0.45267786347581634,
      "grad_norm": 0.18731923231996667,
      "learning_rate": 0.00012005828807417909,
      "loss": 0.9932,
      "step": 6280
    },
    {
      "epoch": 0.45274994593815326,
      "grad_norm": 0.18412421384158736,
      "learning_rate": 0.00012003541486690216,
      "loss": 1.1336,
      "step": 6281
    },
    {
      "epoch": 0.4528220284004902,
      "grad_norm": 0.15977679518921667,
      "learning_rate": 0.00012001254056751587,
      "loss": 1.0719,
      "step": 6282
    },
    {
      "epoch": 0.45289411086282705,
      "grad_norm": 0.19256180890125366,
      "learning_rate": 0.0001199896651772671,
      "loss": 1.0034,
      "step": 6283
    },
    {
      "epoch": 0.45296619332516397,
      "grad_norm": 0.3133401563207344,
      "learning_rate": 0.0001199667886974027,
      "loss": 1.0358,
      "step": 6284
    },
    {
      "epoch": 0.4530382757875009,
      "grad_norm": 0.2863950710822954,
      "learning_rate": 0.00011994391112916977,
      "loss": 1.0727,
      "step": 6285
    },
    {
      "epoch": 0.4531103582498378,
      "grad_norm": 0.18063382222920063,
      "learning_rate": 0.00011992103247381516,
      "loss": 1.1249,
      "step": 6286
    },
    {
      "epoch": 0.45318244071217473,
      "grad_norm": 0.11922656573147722,
      "learning_rate": 0.00011989815273258614,
      "loss": 1.0496,
      "step": 6287
    },
    {
      "epoch": 0.45325452317451165,
      "grad_norm": 0.13870786178346484,
      "learning_rate": 0.00011987527190672973,
      "loss": 1.1078,
      "step": 6288
    },
    {
      "epoch": 0.45332660563684857,
      "grad_norm": 0.13493391808675093,
      "learning_rate": 0.00011985238999749325,
      "loss": 0.9962,
      "step": 6289
    },
    {
      "epoch": 0.4533986880991855,
      "grad_norm": 0.13892737245309703,
      "learning_rate": 0.00011982950700612383,
      "loss": 1.0126,
      "step": 6290
    },
    {
      "epoch": 0.4534707705615224,
      "grad_norm": 0.11959219530455877,
      "learning_rate": 0.00011980662293386896,
      "loss": 0.9753,
      "step": 6291
    },
    {
      "epoch": 0.45354285302385927,
      "grad_norm": 0.12186235157461207,
      "learning_rate": 0.0001197837377819759,
      "loss": 1.0441,
      "step": 6292
    },
    {
      "epoch": 0.4536149354861962,
      "grad_norm": 0.14039958109950226,
      "learning_rate": 0.00011976085155169215,
      "loss": 1.1318,
      "step": 6293
    },
    {
      "epoch": 0.4536870179485331,
      "grad_norm": 0.12130848204597834,
      "learning_rate": 0.0001197379642442652,
      "loss": 1.011,
      "step": 6294
    },
    {
      "epoch": 0.45375910041087003,
      "grad_norm": 0.19022524895545193,
      "learning_rate": 0.00011971507586094261,
      "loss": 1.0022,
      "step": 6295
    },
    {
      "epoch": 0.45383118287320695,
      "grad_norm": 0.14535177188340817,
      "learning_rate": 0.00011969218640297204,
      "loss": 1.1059,
      "step": 6296
    },
    {
      "epoch": 0.45390326533554387,
      "grad_norm": 0.1991234798827582,
      "learning_rate": 0.00011966929587160111,
      "loss": 0.9776,
      "step": 6297
    },
    {
      "epoch": 0.4539753477978808,
      "grad_norm": 0.10198045954531527,
      "learning_rate": 0.0001196464042680776,
      "loss": 1.0257,
      "step": 6298
    },
    {
      "epoch": 0.4540474302602177,
      "grad_norm": 0.13881681151100825,
      "learning_rate": 0.00011962351159364931,
      "loss": 1.0756,
      "step": 6299
    },
    {
      "epoch": 0.45411951272255463,
      "grad_norm": 0.13447103868266838,
      "learning_rate": 0.00011960061784956404,
      "loss": 1.1232,
      "step": 6300
    },
    {
      "epoch": 0.4541915951848915,
      "grad_norm": 0.12055297398366349,
      "learning_rate": 0.00011957772303706978,
      "loss": 1.031,
      "step": 6301
    },
    {
      "epoch": 0.4542636776472284,
      "grad_norm": 0.17875622269265445,
      "learning_rate": 0.00011955482715741446,
      "loss": 1.0567,
      "step": 6302
    },
    {
      "epoch": 0.45433576010956533,
      "grad_norm": 0.16836516736846577,
      "learning_rate": 0.00011953193021184613,
      "loss": 0.9769,
      "step": 6303
    },
    {
      "epoch": 0.45440784257190225,
      "grad_norm": 1.4591455771914665,
      "learning_rate": 0.00011950903220161285,
      "loss": 1.0166,
      "step": 6304
    },
    {
      "epoch": 0.4544799250342392,
      "grad_norm": 0.481240171888939,
      "learning_rate": 0.00011948613312796277,
      "loss": 1.0615,
      "step": 6305
    },
    {
      "epoch": 0.4545520074965761,
      "grad_norm": 0.11730616668494977,
      "learning_rate": 0.00011946323299214412,
      "loss": 1.0898,
      "step": 6306
    },
    {
      "epoch": 0.454624089958913,
      "grad_norm": 0.14028691716965805,
      "learning_rate": 0.00011944033179540515,
      "loss": 1.1041,
      "step": 6307
    },
    {
      "epoch": 0.45469617242124993,
      "grad_norm": 0.1093407304240486,
      "learning_rate": 0.00011941742953899417,
      "loss": 1.082,
      "step": 6308
    },
    {
      "epoch": 0.4547682548835868,
      "grad_norm": 0.17294089860156925,
      "learning_rate": 0.00011939452622415959,
      "loss": 1.0374,
      "step": 6309
    },
    {
      "epoch": 0.4548403373459237,
      "grad_norm": 0.33326472942260227,
      "learning_rate": 0.0001193716218521498,
      "loss": 1.1034,
      "step": 6310
    },
    {
      "epoch": 0.45491241980826064,
      "grad_norm": 0.25260398442172327,
      "learning_rate": 0.00011934871642421333,
      "loss": 1.096,
      "step": 6311
    },
    {
      "epoch": 0.45498450227059756,
      "grad_norm": 0.2484038832949873,
      "learning_rate": 0.0001193258099415987,
      "loss": 1.1499,
      "step": 6312
    },
    {
      "epoch": 0.4550565847329345,
      "grad_norm": 0.13759842043805168,
      "learning_rate": 0.00011930290240555456,
      "loss": 1.0391,
      "step": 6313
    },
    {
      "epoch": 0.4551286671952714,
      "grad_norm": 0.1459809409764024,
      "learning_rate": 0.0001192799938173295,
      "loss": 1.0451,
      "step": 6314
    },
    {
      "epoch": 0.4552007496576083,
      "grad_norm": 0.12263818829390871,
      "learning_rate": 0.00011925708417817234,
      "loss": 1.053,
      "step": 6315
    },
    {
      "epoch": 0.45527283211994524,
      "grad_norm": 0.11911959264816815,
      "learning_rate": 0.00011923417348933183,
      "loss": 1.0211,
      "step": 6316
    },
    {
      "epoch": 0.45534491458228216,
      "grad_norm": 0.12710026312223843,
      "learning_rate": 0.00011921126175205675,
      "loss": 1.1148,
      "step": 6317
    },
    {
      "epoch": 0.455416997044619,
      "grad_norm": 0.1073096845434803,
      "learning_rate": 0.00011918834896759606,
      "loss": 1.1713,
      "step": 6318
    },
    {
      "epoch": 0.45548907950695594,
      "grad_norm": 0.17199703583759662,
      "learning_rate": 0.00011916543513719864,
      "loss": 1.0141,
      "step": 6319
    },
    {
      "epoch": 0.45556116196929286,
      "grad_norm": 0.2860924285736027,
      "learning_rate": 0.00011914252026211361,
      "loss": 1.129,
      "step": 6320
    },
    {
      "epoch": 0.4556332444316298,
      "grad_norm": 0.12551582522373061,
      "learning_rate": 0.00011911960434358991,
      "loss": 1.0774,
      "step": 6321
    },
    {
      "epoch": 0.4557053268939667,
      "grad_norm": 0.36677226457914747,
      "learning_rate": 0.00011909668738287679,
      "loss": 1.166,
      "step": 6322
    },
    {
      "epoch": 0.4557774093563036,
      "grad_norm": 0.19681639631266526,
      "learning_rate": 0.0001190737693812233,
      "loss": 1.0871,
      "step": 6323
    },
    {
      "epoch": 0.45584949181864054,
      "grad_norm": 0.18194612948323408,
      "learning_rate": 0.0001190508503398788,
      "loss": 1.055,
      "step": 6324
    },
    {
      "epoch": 0.45592157428097746,
      "grad_norm": 0.15962900059915988,
      "learning_rate": 0.00011902793026009247,
      "loss": 1.0474,
      "step": 6325
    },
    {
      "epoch": 0.4559936567433143,
      "grad_norm": 0.12056024735683181,
      "learning_rate": 0.00011900500914311376,
      "loss": 1.0306,
      "step": 6326
    },
    {
      "epoch": 0.45606573920565124,
      "grad_norm": 0.11160513775798624,
      "learning_rate": 0.000118982086990192,
      "loss": 1.0851,
      "step": 6327
    },
    {
      "epoch": 0.45613782166798816,
      "grad_norm": 0.12484468648774091,
      "learning_rate": 0.00011895916380257668,
      "loss": 1.0402,
      "step": 6328
    },
    {
      "epoch": 0.4562099041303251,
      "grad_norm": 0.11445552781520346,
      "learning_rate": 0.00011893623958151732,
      "loss": 1.1064,
      "step": 6329
    },
    {
      "epoch": 0.456281986592662,
      "grad_norm": 0.2044575639024178,
      "learning_rate": 0.00011891331432826351,
      "loss": 1.015,
      "step": 6330
    },
    {
      "epoch": 0.4563540690549989,
      "grad_norm": 0.1266572721608095,
      "learning_rate": 0.00011889038804406485,
      "loss": 1.0355,
      "step": 6331
    },
    {
      "epoch": 0.45642615151733584,
      "grad_norm": 0.1262793254118898,
      "learning_rate": 0.00011886746073017105,
      "loss": 1.0748,
      "step": 6332
    },
    {
      "epoch": 0.45649823397967276,
      "grad_norm": 0.1235385255346835,
      "learning_rate": 0.00011884453238783185,
      "loss": 1.0348,
      "step": 6333
    },
    {
      "epoch": 0.4565703164420097,
      "grad_norm": 0.12232167127474178,
      "learning_rate": 0.00011882160301829703,
      "loss": 0.9802,
      "step": 6334
    },
    {
      "epoch": 0.45664239890434655,
      "grad_norm": 0.12434786297942078,
      "learning_rate": 0.00011879867262281648,
      "loss": 1.0281,
      "step": 6335
    },
    {
      "epoch": 0.45671448136668347,
      "grad_norm": 0.11156210337743806,
      "learning_rate": 0.00011877574120264008,
      "loss": 1.0129,
      "step": 6336
    },
    {
      "epoch": 0.4567865638290204,
      "grad_norm": 0.15967270294580832,
      "learning_rate": 0.00011875280875901782,
      "loss": 1.014,
      "step": 6337
    },
    {
      "epoch": 0.4568586462913573,
      "grad_norm": 0.17235756665573998,
      "learning_rate": 0.00011872987529319974,
      "loss": 1.1057,
      "step": 6338
    },
    {
      "epoch": 0.4569307287536942,
      "grad_norm": 0.16392501106378166,
      "learning_rate": 0.00011870694080643585,
      "loss": 1.0515,
      "step": 6339
    },
    {
      "epoch": 0.45700281121603115,
      "grad_norm": 0.15003262005522447,
      "learning_rate": 0.00011868400529997636,
      "loss": 1.0281,
      "step": 6340
    },
    {
      "epoch": 0.45707489367836807,
      "grad_norm": 0.1152054211683676,
      "learning_rate": 0.0001186610687750714,
      "loss": 1.0073,
      "step": 6341
    },
    {
      "epoch": 0.457146976140705,
      "grad_norm": 0.13769332030668308,
      "learning_rate": 0.00011863813123297128,
      "loss": 1.0492,
      "step": 6342
    },
    {
      "epoch": 0.4572190586030419,
      "grad_norm": 0.12363109936235915,
      "learning_rate": 0.00011861519267492623,
      "loss": 0.9593,
      "step": 6343
    },
    {
      "epoch": 0.45729114106537877,
      "grad_norm": 0.26245537379072087,
      "learning_rate": 0.00011859225310218666,
      "loss": 1.1191,
      "step": 6344
    },
    {
      "epoch": 0.4573632235277157,
      "grad_norm": 0.12516116025197283,
      "learning_rate": 0.00011856931251600297,
      "loss": 1.0493,
      "step": 6345
    },
    {
      "epoch": 0.4574353059900526,
      "grad_norm": 0.32333201276092616,
      "learning_rate": 0.00011854637091762563,
      "loss": 1.0373,
      "step": 6346
    },
    {
      "epoch": 0.45750738845238953,
      "grad_norm": 0.1256097827442289,
      "learning_rate": 0.00011852342830830511,
      "loss": 1.0361,
      "step": 6347
    },
    {
      "epoch": 0.45757947091472645,
      "grad_norm": 0.12896877621481323,
      "learning_rate": 0.00011850048468929209,
      "loss": 1.0913,
      "step": 6348
    },
    {
      "epoch": 0.45765155337706337,
      "grad_norm": 0.1251115748730228,
      "learning_rate": 0.00011847754006183713,
      "loss": 1.0213,
      "step": 6349
    },
    {
      "epoch": 0.4577236358394003,
      "grad_norm": 0.18079384240365054,
      "learning_rate": 0.00011845459442719092,
      "loss": 1.0015,
      "step": 6350
    },
    {
      "epoch": 0.4577957183017372,
      "grad_norm": 0.23095765674970828,
      "learning_rate": 0.0001184316477866042,
      "loss": 1.0476,
      "step": 6351
    },
    {
      "epoch": 0.4578678007640741,
      "grad_norm": 0.1436908826179935,
      "learning_rate": 0.00011840870014132782,
      "loss": 1.0656,
      "step": 6352
    },
    {
      "epoch": 0.457939883226411,
      "grad_norm": 0.12037565121054743,
      "learning_rate": 0.00011838575149261255,
      "loss": 1.0225,
      "step": 6353
    },
    {
      "epoch": 0.4580119656887479,
      "grad_norm": 0.14530408823989968,
      "learning_rate": 0.00011836280184170938,
      "loss": 1.0433,
      "step": 6354
    },
    {
      "epoch": 0.45808404815108483,
      "grad_norm": 0.17290327666201483,
      "learning_rate": 0.00011833985118986919,
      "loss": 1.1178,
      "step": 6355
    },
    {
      "epoch": 0.45815613061342175,
      "grad_norm": 0.18216144298388176,
      "learning_rate": 0.00011831689953834306,
      "loss": 1.0301,
      "step": 6356
    },
    {
      "epoch": 0.4582282130757587,
      "grad_norm": 0.11920127226470337,
      "learning_rate": 0.00011829394688838205,
      "loss": 0.9555,
      "step": 6357
    },
    {
      "epoch": 0.4583002955380956,
      "grad_norm": 0.12945548988341066,
      "learning_rate": 0.00011827099324123725,
      "loss": 1.1076,
      "step": 6358
    },
    {
      "epoch": 0.4583723780004325,
      "grad_norm": 0.11289452638836939,
      "learning_rate": 0.00011824803859815988,
      "loss": 1.0186,
      "step": 6359
    },
    {
      "epoch": 0.45844446046276943,
      "grad_norm": 0.10941143549845822,
      "learning_rate": 0.00011822508296040116,
      "loss": 1.1012,
      "step": 6360
    },
    {
      "epoch": 0.4585165429251063,
      "grad_norm": 0.12207155218266838,
      "learning_rate": 0.00011820212632921233,
      "loss": 1.0033,
      "step": 6361
    },
    {
      "epoch": 0.4585886253874432,
      "grad_norm": 0.13420346823445306,
      "learning_rate": 0.00011817916870584482,
      "loss": 1.082,
      "step": 6362
    },
    {
      "epoch": 0.45866070784978014,
      "grad_norm": 0.13510534055703577,
      "learning_rate": 0.00011815621009154994,
      "loss": 1.0199,
      "step": 6363
    },
    {
      "epoch": 0.45873279031211706,
      "grad_norm": 0.17307480225659388,
      "learning_rate": 0.00011813325048757922,
      "loss": 1.038,
      "step": 6364
    },
    {
      "epoch": 0.458804872774454,
      "grad_norm": 0.11539348808632109,
      "learning_rate": 0.00011811028989518409,
      "loss": 1.0294,
      "step": 6365
    },
    {
      "epoch": 0.4588769552367909,
      "grad_norm": 0.20238244765374852,
      "learning_rate": 0.00011808732831561617,
      "loss": 1.0754,
      "step": 6366
    },
    {
      "epoch": 0.4589490376991278,
      "grad_norm": 0.11406883547516514,
      "learning_rate": 0.000118064365750127,
      "loss": 0.9821,
      "step": 6367
    },
    {
      "epoch": 0.45902112016146474,
      "grad_norm": 0.20087124894251962,
      "learning_rate": 0.00011804140219996835,
      "loss": 1.0115,
      "step": 6368
    },
    {
      "epoch": 0.45909320262380166,
      "grad_norm": 0.20257851870255486,
      "learning_rate": 0.00011801843766639182,
      "loss": 1.0087,
      "step": 6369
    },
    {
      "epoch": 0.4591652850861385,
      "grad_norm": 0.11827222763623685,
      "learning_rate": 0.00011799547215064928,
      "loss": 1.142,
      "step": 6370
    },
    {
      "epoch": 0.45923736754847544,
      "grad_norm": 0.15311298733944656,
      "learning_rate": 0.00011797250565399252,
      "loss": 1.1646,
      "step": 6371
    },
    {
      "epoch": 0.45930945001081236,
      "grad_norm": 0.11621280208083042,
      "learning_rate": 0.00011794953817767338,
      "loss": 1.0418,
      "step": 6372
    },
    {
      "epoch": 0.4593815324731493,
      "grad_norm": 0.11953311826111318,
      "learning_rate": 0.00011792656972294385,
      "loss": 1.0059,
      "step": 6373
    },
    {
      "epoch": 0.4594536149354862,
      "grad_norm": 0.13070686429427153,
      "learning_rate": 0.00011790360029105588,
      "loss": 1.071,
      "step": 6374
    },
    {
      "epoch": 0.4595256973978231,
      "grad_norm": 0.16839234791350344,
      "learning_rate": 0.00011788062988326155,
      "loss": 1.1141,
      "step": 6375
    },
    {
      "epoch": 0.45959777986016004,
      "grad_norm": 0.12207941986069429,
      "learning_rate": 0.0001178576585008129,
      "loss": 1.0703,
      "step": 6376
    },
    {
      "epoch": 0.45966986232249696,
      "grad_norm": 0.11451243894386287,
      "learning_rate": 0.00011783468614496209,
      "loss": 1.028,
      "step": 6377
    },
    {
      "epoch": 0.4597419447848338,
      "grad_norm": 0.11414556931833733,
      "learning_rate": 0.00011781171281696135,
      "loss": 1.0696,
      "step": 6378
    },
    {
      "epoch": 0.45981402724717074,
      "grad_norm": 0.15810524232590642,
      "learning_rate": 0.00011778873851806294,
      "loss": 1.1246,
      "step": 6379
    },
    {
      "epoch": 0.45988610970950766,
      "grad_norm": 0.1896863291481581,
      "learning_rate": 0.0001177657632495191,
      "loss": 1.106,
      "step": 6380
    },
    {
      "epoch": 0.4599581921718446,
      "grad_norm": 0.12467544672176245,
      "learning_rate": 0.00011774278701258223,
      "loss": 1.0281,
      "step": 6381
    },
    {
      "epoch": 0.4600302746341815,
      "grad_norm": 0.11537226987894579,
      "learning_rate": 0.00011771980980850475,
      "loss": 1.1755,
      "step": 6382
    },
    {
      "epoch": 0.4601023570965184,
      "grad_norm": 0.11054844401910634,
      "learning_rate": 0.0001176968316385391,
      "loss": 1.0932,
      "step": 6383
    },
    {
      "epoch": 0.46017443955885534,
      "grad_norm": 0.11740960159725045,
      "learning_rate": 0.00011767385250393781,
      "loss": 1.0626,
      "step": 6384
    },
    {
      "epoch": 0.46024652202119226,
      "grad_norm": 0.12199344425459256,
      "learning_rate": 0.00011765087240595345,
      "loss": 1.1025,
      "step": 6385
    },
    {
      "epoch": 0.4603186044835292,
      "grad_norm": 0.12194013167138763,
      "learning_rate": 0.00011762789134583863,
      "loss": 1.0867,
      "step": 6386
    },
    {
      "epoch": 0.46039068694586605,
      "grad_norm": 0.11375595111045154,
      "learning_rate": 0.00011760490932484602,
      "loss": 1.0797,
      "step": 6387
    },
    {
      "epoch": 0.46046276940820297,
      "grad_norm": 0.1242669749765249,
      "learning_rate": 0.00011758192634422837,
      "loss": 1.0263,
      "step": 6388
    },
    {
      "epoch": 0.4605348518705399,
      "grad_norm": 0.156089177103708,
      "learning_rate": 0.00011755894240523844,
      "loss": 1.0907,
      "step": 6389
    },
    {
      "epoch": 0.4606069343328768,
      "grad_norm": 0.1301720209647431,
      "learning_rate": 0.00011753595750912905,
      "loss": 1.0257,
      "step": 6390
    },
    {
      "epoch": 0.4606790167952137,
      "grad_norm": 0.13797070540516185,
      "learning_rate": 0.00011751297165715309,
      "loss": 1.0584,
      "step": 6391
    },
    {
      "epoch": 0.46075109925755064,
      "grad_norm": 0.11096220085963432,
      "learning_rate": 0.0001174899848505635,
      "loss": 0.9973,
      "step": 6392
    },
    {
      "epoch": 0.46082318171988756,
      "grad_norm": 0.14748787884387024,
      "learning_rate": 0.00011746699709061331,
      "loss": 1.112,
      "step": 6393
    },
    {
      "epoch": 0.4608952641822245,
      "grad_norm": 0.1257892661636559,
      "learning_rate": 0.00011744400837855545,
      "loss": 1.0765,
      "step": 6394
    },
    {
      "epoch": 0.4609673466445614,
      "grad_norm": 0.14735201821647587,
      "learning_rate": 0.00011742101871564313,
      "loss": 1.1046,
      "step": 6395
    },
    {
      "epoch": 0.46103942910689827,
      "grad_norm": 0.12014049025851388,
      "learning_rate": 0.00011739802810312939,
      "loss": 1.0651,
      "step": 6396
    },
    {
      "epoch": 0.4611115115692352,
      "grad_norm": 0.14102389996797043,
      "learning_rate": 0.00011737503654226751,
      "loss": 1.1127,
      "step": 6397
    },
    {
      "epoch": 0.4611835940315721,
      "grad_norm": 0.10925272284888034,
      "learning_rate": 0.00011735204403431066,
      "loss": 1.0195,
      "step": 6398
    },
    {
      "epoch": 0.46125567649390903,
      "grad_norm": 0.1172059636293125,
      "learning_rate": 0.0001173290505805122,
      "loss": 1.0504,
      "step": 6399
    },
    {
      "epoch": 0.46132775895624595,
      "grad_norm": 0.1452243174005221,
      "learning_rate": 0.00011730605618212544,
      "loss": 0.9784,
      "step": 6400
    },
    {
      "epoch": 0.46139984141858287,
      "grad_norm": 0.13049552665201677,
      "learning_rate": 0.00011728306084040381,
      "loss": 1.018,
      "step": 6401
    },
    {
      "epoch": 0.4614719238809198,
      "grad_norm": 0.13498742325129398,
      "learning_rate": 0.00011726006455660072,
      "loss": 1.083,
      "step": 6402
    },
    {
      "epoch": 0.4615440063432567,
      "grad_norm": 0.1795179384133058,
      "learning_rate": 0.00011723706733196973,
      "loss": 1.0383,
      "step": 6403
    },
    {
      "epoch": 0.46161608880559357,
      "grad_norm": 0.10940398992889955,
      "learning_rate": 0.00011721406916776434,
      "loss": 1.0191,
      "step": 6404
    },
    {
      "epoch": 0.4616881712679305,
      "grad_norm": 0.1528867121868048,
      "learning_rate": 0.0001171910700652382,
      "loss": 1.0163,
      "step": 6405
    },
    {
      "epoch": 0.4617602537302674,
      "grad_norm": 0.12256389908137967,
      "learning_rate": 0.00011716807002564491,
      "loss": 1.0673,
      "step": 6406
    },
    {
      "epoch": 0.46183233619260433,
      "grad_norm": 0.11705892156297792,
      "learning_rate": 0.00011714506905023826,
      "loss": 1.0873,
      "step": 6407
    },
    {
      "epoch": 0.46190441865494125,
      "grad_norm": 0.24875882430321014,
      "learning_rate": 0.00011712206714027192,
      "loss": 1.072,
      "step": 6408
    },
    {
      "epoch": 0.46197650111727817,
      "grad_norm": 0.12582601901287643,
      "learning_rate": 0.00011709906429699977,
      "loss": 1.0251,
      "step": 6409
    },
    {
      "epoch": 0.4620485835796151,
      "grad_norm": 0.11272745850962343,
      "learning_rate": 0.00011707606052167563,
      "loss": 1.0664,
      "step": 6410
    },
    {
      "epoch": 0.462120666041952,
      "grad_norm": 0.12147416683698999,
      "learning_rate": 0.00011705305581555344,
      "loss": 1.1139,
      "step": 6411
    },
    {
      "epoch": 0.46219274850428893,
      "grad_norm": 0.17770771415508732,
      "learning_rate": 0.00011703005017988712,
      "loss": 1.0472,
      "step": 6412
    },
    {
      "epoch": 0.4622648309666258,
      "grad_norm": 0.12268375431981626,
      "learning_rate": 0.00011700704361593073,
      "loss": 1.0394,
      "step": 6413
    },
    {
      "epoch": 0.4623369134289627,
      "grad_norm": 0.20518384655769772,
      "learning_rate": 0.00011698403612493832,
      "loss": 1.0699,
      "step": 6414
    },
    {
      "epoch": 0.46240899589129963,
      "grad_norm": 0.11579932746550846,
      "learning_rate": 0.00011696102770816399,
      "loss": 0.9875,
      "step": 6415
    },
    {
      "epoch": 0.46248107835363655,
      "grad_norm": 0.11740707954064998,
      "learning_rate": 0.00011693801836686189,
      "loss": 1.1017,
      "step": 6416
    },
    {
      "epoch": 0.4625531608159735,
      "grad_norm": 0.149302945428576,
      "learning_rate": 0.0001169150081022863,
      "loss": 1.0137,
      "step": 6417
    },
    {
      "epoch": 0.4626252432783104,
      "grad_norm": 0.12771221365553911,
      "learning_rate": 0.0001168919969156914,
      "loss": 1.1028,
      "step": 6418
    },
    {
      "epoch": 0.4626973257406473,
      "grad_norm": 0.15622969830907796,
      "learning_rate": 0.0001168689848083316,
      "loss": 1.0544,
      "step": 6419
    },
    {
      "epoch": 0.46276940820298423,
      "grad_norm": 0.11127131998746344,
      "learning_rate": 0.00011684597178146115,
      "loss": 1.0473,
      "step": 6420
    },
    {
      "epoch": 0.46284149066532115,
      "grad_norm": 0.678882821028855,
      "learning_rate": 0.00011682295783633457,
      "loss": 1.0665,
      "step": 6421
    },
    {
      "epoch": 0.462913573127658,
      "grad_norm": 0.14291997164986608,
      "learning_rate": 0.00011679994297420627,
      "loss": 1.1177,
      "step": 6422
    },
    {
      "epoch": 0.46298565558999494,
      "grad_norm": 0.11316098348607319,
      "learning_rate": 0.0001167769271963308,
      "loss": 1.0481,
      "step": 6423
    },
    {
      "epoch": 0.46305773805233186,
      "grad_norm": 0.12251597143461065,
      "learning_rate": 0.00011675391050396267,
      "loss": 1.1024,
      "step": 6424
    },
    {
      "epoch": 0.4631298205146688,
      "grad_norm": 0.34032101772275497,
      "learning_rate": 0.00011673089289835657,
      "loss": 1.0686,
      "step": 6425
    },
    {
      "epoch": 0.4632019029770057,
      "grad_norm": 0.12053602407881606,
      "learning_rate": 0.00011670787438076711,
      "loss": 1.089,
      "step": 6426
    },
    {
      "epoch": 0.4632739854393426,
      "grad_norm": 0.15320510444176813,
      "learning_rate": 0.00011668485495244903,
      "loss": 1.1047,
      "step": 6427
    },
    {
      "epoch": 0.46334606790167954,
      "grad_norm": 0.14301692723781426,
      "learning_rate": 0.00011666183461465709,
      "loss": 0.9674,
      "step": 6428
    },
    {
      "epoch": 0.46341815036401646,
      "grad_norm": 0.12224607354716517,
      "learning_rate": 0.00011663881336864611,
      "loss": 1.0481,
      "step": 6429
    },
    {
      "epoch": 0.4634902328263533,
      "grad_norm": 0.11643077075416437,
      "learning_rate": 0.00011661579121567092,
      "loss": 1.0884,
      "step": 6430
    },
    {
      "epoch": 0.46356231528869024,
      "grad_norm": 0.29587230781206963,
      "learning_rate": 0.00011659276815698648,
      "loss": 1.056,
      "step": 6431
    },
    {
      "epoch": 0.46363439775102716,
      "grad_norm": 0.10757924204175653,
      "learning_rate": 0.00011656974419384771,
      "loss": 1.0549,
      "step": 6432
    },
    {
      "epoch": 0.4637064802133641,
      "grad_norm": 0.12963722708153036,
      "learning_rate": 0.00011654671932750969,
      "loss": 1.1015,
      "step": 6433
    },
    {
      "epoch": 0.463778562675701,
      "grad_norm": 0.12874242261530724,
      "learning_rate": 0.00011652369355922739,
      "loss": 1.0628,
      "step": 6434
    },
    {
      "epoch": 0.4638506451380379,
      "grad_norm": 0.11311021323542587,
      "learning_rate": 0.00011650066689025599,
      "loss": 0.9871,
      "step": 6435
    },
    {
      "epoch": 0.46392272760037484,
      "grad_norm": 0.13626564568794655,
      "learning_rate": 0.0001164776393218506,
      "loss": 1.063,
      "step": 6436
    },
    {
      "epoch": 0.46399481006271176,
      "grad_norm": 0.19762916668356137,
      "learning_rate": 0.0001164546108552665,
      "loss": 1.0431,
      "step": 6437
    },
    {
      "epoch": 0.4640668925250487,
      "grad_norm": 0.12351601583438578,
      "learning_rate": 0.00011643158149175886,
      "loss": 0.9811,
      "step": 6438
    },
    {
      "epoch": 0.46413897498738554,
      "grad_norm": 0.1859810821402429,
      "learning_rate": 0.00011640855123258306,
      "loss": 1.0116,
      "step": 6439
    },
    {
      "epoch": 0.46421105744972246,
      "grad_norm": 0.13863098377775834,
      "learning_rate": 0.00011638552007899437,
      "loss": 1.019,
      "step": 6440
    },
    {
      "epoch": 0.4642831399120594,
      "grad_norm": 0.1458451122416303,
      "learning_rate": 0.00011636248803224832,
      "loss": 1.061,
      "step": 6441
    },
    {
      "epoch": 0.4643552223743963,
      "grad_norm": 0.12495409498452185,
      "learning_rate": 0.00011633945509360023,
      "loss": 1.1132,
      "step": 6442
    },
    {
      "epoch": 0.4644273048367332,
      "grad_norm": 0.20016852369749952,
      "learning_rate": 0.00011631642126430568,
      "loss": 1.0843,
      "step": 6443
    },
    {
      "epoch": 0.46449938729907014,
      "grad_norm": 0.13623705127775287,
      "learning_rate": 0.00011629338654562018,
      "loss": 1.03,
      "step": 6444
    },
    {
      "epoch": 0.46457146976140706,
      "grad_norm": 0.11814792639835225,
      "learning_rate": 0.00011627035093879939,
      "loss": 1.0215,
      "step": 6445
    },
    {
      "epoch": 0.464643552223744,
      "grad_norm": 0.2818458085707111,
      "learning_rate": 0.0001162473144450989,
      "loss": 1.1635,
      "step": 6446
    },
    {
      "epoch": 0.4647156346860809,
      "grad_norm": 0.1166716984243,
      "learning_rate": 0.00011622427706577441,
      "loss": 1.0828,
      "step": 6447
    },
    {
      "epoch": 0.46478771714841777,
      "grad_norm": 0.11710357608219288,
      "learning_rate": 0.00011620123880208167,
      "loss": 1.0777,
      "step": 6448
    },
    {
      "epoch": 0.4648597996107547,
      "grad_norm": 0.20695250991880118,
      "learning_rate": 0.0001161781996552765,
      "loss": 1.1434,
      "step": 6449
    },
    {
      "epoch": 0.4649318820730916,
      "grad_norm": 0.1166594900601599,
      "learning_rate": 0.00011615515962661469,
      "loss": 1.0681,
      "step": 6450
    },
    {
      "epoch": 0.4650039645354285,
      "grad_norm": 0.11166240924500541,
      "learning_rate": 0.00011613211871735214,
      "loss": 0.9522,
      "step": 6451
    },
    {
      "epoch": 0.46507604699776545,
      "grad_norm": 0.11645000225801705,
      "learning_rate": 0.00011610907692874482,
      "loss": 1.0586,
      "step": 6452
    },
    {
      "epoch": 0.46514812946010237,
      "grad_norm": 0.10675177365643307,
      "learning_rate": 0.0001160860342620487,
      "loss": 1.0576,
      "step": 6453
    },
    {
      "epoch": 0.4652202119224393,
      "grad_norm": 0.11330605511153997,
      "learning_rate": 0.00011606299071851978,
      "loss": 1.0467,
      "step": 6454
    },
    {
      "epoch": 0.4652922943847762,
      "grad_norm": 0.12076331459221121,
      "learning_rate": 0.00011603994629941417,
      "loss": 1.069,
      "step": 6455
    },
    {
      "epoch": 0.46536437684711307,
      "grad_norm": 0.12352482006290307,
      "learning_rate": 0.00011601690100598801,
      "loss": 1.1133,
      "step": 6456
    },
    {
      "epoch": 0.46543645930945,
      "grad_norm": 0.13635243684669202,
      "learning_rate": 0.00011599385483949743,
      "loss": 1.0644,
      "step": 6457
    },
    {
      "epoch": 0.4655085417717869,
      "grad_norm": 0.12562698110866596,
      "learning_rate": 0.00011597080780119868,
      "loss": 1.0267,
      "step": 6458
    },
    {
      "epoch": 0.46558062423412383,
      "grad_norm": 0.10274855565099274,
      "learning_rate": 0.00011594775989234803,
      "loss": 1.0735,
      "step": 6459
    },
    {
      "epoch": 0.46565270669646075,
      "grad_norm": 0.15071377192466887,
      "learning_rate": 0.00011592471111420178,
      "loss": 1.0734,
      "step": 6460
    },
    {
      "epoch": 0.46572478915879767,
      "grad_norm": 0.10446268829502334,
      "learning_rate": 0.00011590166146801633,
      "loss": 1.0094,
      "step": 6461
    },
    {
      "epoch": 0.4657968716211346,
      "grad_norm": 0.14022236363963622,
      "learning_rate": 0.00011587861095504805,
      "loss": 1.0482,
      "step": 6462
    },
    {
      "epoch": 0.4658689540834715,
      "grad_norm": 0.14510006410904297,
      "learning_rate": 0.00011585555957655342,
      "loss": 1.0294,
      "step": 6463
    },
    {
      "epoch": 0.46594103654580843,
      "grad_norm": 0.12724511696194377,
      "learning_rate": 0.00011583250733378896,
      "loss": 1.0428,
      "step": 6464
    },
    {
      "epoch": 0.4660131190081453,
      "grad_norm": 0.12351904470918054,
      "learning_rate": 0.00011580945422801119,
      "loss": 1.0461,
      "step": 6465
    },
    {
      "epoch": 0.4660852014704822,
      "grad_norm": 0.14749851384728782,
      "learning_rate": 0.00011578640026047676,
      "loss": 1.0558,
      "step": 6466
    },
    {
      "epoch": 0.46615728393281913,
      "grad_norm": 0.119407070527178,
      "learning_rate": 0.00011576334543244225,
      "loss": 1.0462,
      "step": 6467
    },
    {
      "epoch": 0.46622936639515605,
      "grad_norm": 0.1548478252410011,
      "learning_rate": 0.00011574028974516444,
      "loss": 1.1233,
      "step": 6468
    },
    {
      "epoch": 0.466301448857493,
      "grad_norm": 0.17408038203388163,
      "learning_rate": 0.00011571723319989996,
      "loss": 1.0183,
      "step": 6469
    },
    {
      "epoch": 0.4663735313198299,
      "grad_norm": 0.16367218629627728,
      "learning_rate": 0.00011569417579790571,
      "loss": 1.0514,
      "step": 6470
    },
    {
      "epoch": 0.4664456137821668,
      "grad_norm": 0.660205283950124,
      "learning_rate": 0.00011567111754043844,
      "loss": 1.0397,
      "step": 6471
    },
    {
      "epoch": 0.46651769624450373,
      "grad_norm": 0.11223681699487979,
      "learning_rate": 0.00011564805842875511,
      "loss": 1.1861,
      "step": 6472
    },
    {
      "epoch": 0.46658977870684065,
      "grad_norm": 0.1513515063934548,
      "learning_rate": 0.00011562499846411254,
      "loss": 0.9533,
      "step": 6473
    },
    {
      "epoch": 0.4666618611691775,
      "grad_norm": 0.11092036760805528,
      "learning_rate": 0.00011560193764776783,
      "loss": 1.1056,
      "step": 6474
    },
    {
      "epoch": 0.46673394363151444,
      "grad_norm": 0.11689983961897668,
      "learning_rate": 0.0001155788759809779,
      "loss": 1.0215,
      "step": 6475
    },
    {
      "epoch": 0.46680602609385136,
      "grad_norm": 0.14987138236536832,
      "learning_rate": 0.00011555581346499988,
      "loss": 1.0532,
      "step": 6476
    },
    {
      "epoch": 0.4668781085561883,
      "grad_norm": 0.32826900840541506,
      "learning_rate": 0.00011553275010109084,
      "loss": 1.0146,
      "step": 6477
    },
    {
      "epoch": 0.4669501910185252,
      "grad_norm": 0.11923859206882736,
      "learning_rate": 0.00011550968589050799,
      "loss": 1.0265,
      "step": 6478
    },
    {
      "epoch": 0.4670222734808621,
      "grad_norm": 0.12039153161185157,
      "learning_rate": 0.00011548662083450848,
      "loss": 1.0786,
      "step": 6479
    },
    {
      "epoch": 0.46709435594319904,
      "grad_norm": 0.1170672159013447,
      "learning_rate": 0.00011546355493434959,
      "loss": 0.9886,
      "step": 6480
    },
    {
      "epoch": 0.46716643840553596,
      "grad_norm": 0.11983542440395656,
      "learning_rate": 0.00011544048819128862,
      "loss": 1.1158,
      "step": 6481
    },
    {
      "epoch": 0.4672385208678728,
      "grad_norm": 0.150602634712332,
      "learning_rate": 0.0001154174206065829,
      "loss": 1.03,
      "step": 6482
    },
    {
      "epoch": 0.46731060333020974,
      "grad_norm": 0.13294765914268447,
      "learning_rate": 0.00011539435218148984,
      "loss": 1.0157,
      "step": 6483
    },
    {
      "epoch": 0.46738268579254666,
      "grad_norm": 0.1361905978178888,
      "learning_rate": 0.00011537128291726685,
      "loss": 1.0959,
      "step": 6484
    },
    {
      "epoch": 0.4674547682548836,
      "grad_norm": 0.1350876420214284,
      "learning_rate": 0.00011534821281517143,
      "loss": 0.9788,
      "step": 6485
    },
    {
      "epoch": 0.4675268507172205,
      "grad_norm": 0.12929674062342725,
      "learning_rate": 0.00011532514187646111,
      "loss": 1.0197,
      "step": 6486
    },
    {
      "epoch": 0.4675989331795574,
      "grad_norm": 0.12834194175166128,
      "learning_rate": 0.00011530207010239344,
      "loss": 1.103,
      "step": 6487
    },
    {
      "epoch": 0.46767101564189434,
      "grad_norm": 0.12758330861460968,
      "learning_rate": 0.00011527899749422608,
      "loss": 1.0342,
      "step": 6488
    },
    {
      "epoch": 0.46774309810423126,
      "grad_norm": 0.1271054347795652,
      "learning_rate": 0.00011525592405321666,
      "loss": 1.0615,
      "step": 6489
    },
    {
      "epoch": 0.4678151805665682,
      "grad_norm": 0.11280467593230052,
      "learning_rate": 0.0001152328497806229,
      "loss": 1.0514,
      "step": 6490
    },
    {
      "epoch": 0.46788726302890504,
      "grad_norm": 0.1148906563967108,
      "learning_rate": 0.00011520977467770251,
      "loss": 1.116,
      "step": 6491
    },
    {
      "epoch": 0.46795934549124196,
      "grad_norm": 1.43490211757946,
      "learning_rate": 0.0001151866987457134,
      "loss": 1.0019,
      "step": 6492
    },
    {
      "epoch": 0.4680314279535789,
      "grad_norm": 0.12562733195587958,
      "learning_rate": 0.00011516362198591326,
      "loss": 0.999,
      "step": 6493
    },
    {
      "epoch": 0.4681035104159158,
      "grad_norm": 0.11402809390267218,
      "learning_rate": 0.00011514054439956013,
      "loss": 1.0223,
      "step": 6494
    },
    {
      "epoch": 0.4681755928782527,
      "grad_norm": 0.2415984010937761,
      "learning_rate": 0.00011511746598791182,
      "loss": 1.074,
      "step": 6495
    },
    {
      "epoch": 0.46824767534058964,
      "grad_norm": 0.10491983931953151,
      "learning_rate": 0.00011509438675222643,
      "loss": 1.0076,
      "step": 6496
    },
    {
      "epoch": 0.46831975780292656,
      "grad_norm": 0.11629226524019726,
      "learning_rate": 0.00011507130669376187,
      "loss": 1.0117,
      "step": 6497
    },
    {
      "epoch": 0.4683918402652635,
      "grad_norm": 0.12925134533814386,
      "learning_rate": 0.00011504822581377633,
      "loss": 1.012,
      "step": 6498
    },
    {
      "epoch": 0.4684639227276004,
      "grad_norm": 0.1167856602937204,
      "learning_rate": 0.00011502514411352778,
      "loss": 1.0482,
      "step": 6499
    },
    {
      "epoch": 0.46853600518993727,
      "grad_norm": 0.12739308550462244,
      "learning_rate": 0.00011500206159427453,
      "loss": 1.0877,
      "step": 6500
    },
    {
      "epoch": 0.4686080876522742,
      "grad_norm": 0.11362854930416434,
      "learning_rate": 0.00011497897825727469,
      "loss": 1.1145,
      "step": 6501
    },
    {
      "epoch": 0.4686801701146111,
      "grad_norm": 0.2007841060762036,
      "learning_rate": 0.0001149558941037865,
      "loss": 1.074,
      "step": 6502
    },
    {
      "epoch": 0.468752252576948,
      "grad_norm": 0.21145412876673714,
      "learning_rate": 0.0001149328091350683,
      "loss": 1.0252,
      "step": 6503
    },
    {
      "epoch": 0.46882433503928495,
      "grad_norm": 0.1380014043147834,
      "learning_rate": 0.00011490972335237842,
      "loss": 0.961,
      "step": 6504
    },
    {
      "epoch": 0.46889641750162186,
      "grad_norm": 0.1251447732744197,
      "learning_rate": 0.0001148866367569752,
      "loss": 1.0039,
      "step": 6505
    },
    {
      "epoch": 0.4689684999639588,
      "grad_norm": 0.14571725954308587,
      "learning_rate": 0.00011486354935011711,
      "loss": 1.1226,
      "step": 6506
    },
    {
      "epoch": 0.4690405824262957,
      "grad_norm": 0.17733755819385894,
      "learning_rate": 0.00011484046113306262,
      "loss": 0.996,
      "step": 6507
    },
    {
      "epoch": 0.46911266488863257,
      "grad_norm": 0.12351212566821716,
      "learning_rate": 0.00011481737210707022,
      "loss": 1.0511,
      "step": 6508
    },
    {
      "epoch": 0.4691847473509695,
      "grad_norm": 0.12641160528006457,
      "learning_rate": 0.0001147942822733985,
      "loss": 1.0182,
      "step": 6509
    },
    {
      "epoch": 0.4692568298133064,
      "grad_norm": 0.18697852306580037,
      "learning_rate": 0.000114771191633306,
      "loss": 1.073,
      "step": 6510
    },
    {
      "epoch": 0.46932891227564333,
      "grad_norm": 0.744936557176816,
      "learning_rate": 0.00011474810018805144,
      "loss": 1.1282,
      "step": 6511
    },
    {
      "epoch": 0.46940099473798025,
      "grad_norm": 0.10826210974975127,
      "learning_rate": 0.00011472500793889352,
      "loss": 1.0806,
      "step": 6512
    },
    {
      "epoch": 0.46947307720031717,
      "grad_norm": 0.11540456351599084,
      "learning_rate": 0.00011470191488709086,
      "loss": 1.0778,
      "step": 6513
    },
    {
      "epoch": 0.4695451596626541,
      "grad_norm": 0.13895280496519125,
      "learning_rate": 0.00011467882103390235,
      "loss": 1.0669,
      "step": 6514
    },
    {
      "epoch": 0.469617242124991,
      "grad_norm": 0.14708145844675746,
      "learning_rate": 0.00011465572638058675,
      "loss": 1.0351,
      "step": 6515
    },
    {
      "epoch": 0.4696893245873279,
      "grad_norm": 0.6628883188666188,
      "learning_rate": 0.000114632630928403,
      "loss": 1.0559,
      "step": 6516
    },
    {
      "epoch": 0.4697614070496648,
      "grad_norm": 0.18376491911650733,
      "learning_rate": 0.00011460953467860989,
      "loss": 0.9929,
      "step": 6517
    },
    {
      "epoch": 0.4698334895120017,
      "grad_norm": 0.15393376807396397,
      "learning_rate": 0.00011458643763246648,
      "loss": 1.13,
      "step": 6518
    },
    {
      "epoch": 0.46990557197433863,
      "grad_norm": 0.11098602709813686,
      "learning_rate": 0.0001145633397912317,
      "loss": 0.9921,
      "step": 6519
    },
    {
      "epoch": 0.46997765443667555,
      "grad_norm": 0.11725826160594084,
      "learning_rate": 0.00011454024115616464,
      "loss": 1.0704,
      "step": 6520
    },
    {
      "epoch": 0.47004973689901247,
      "grad_norm": 0.1125337370870447,
      "learning_rate": 0.00011451714172852433,
      "loss": 0.9359,
      "step": 6521
    },
    {
      "epoch": 0.4701218193613494,
      "grad_norm": 0.11803951727702805,
      "learning_rate": 0.00011449404150956994,
      "loss": 1.0064,
      "step": 6522
    },
    {
      "epoch": 0.4701939018236863,
      "grad_norm": 0.16084854520578057,
      "learning_rate": 0.00011447094050056064,
      "loss": 1.0102,
      "step": 6523
    },
    {
      "epoch": 0.47026598428602323,
      "grad_norm": 0.22642309251552972,
      "learning_rate": 0.0001144478387027556,
      "loss": 1.1544,
      "step": 6524
    },
    {
      "epoch": 0.47033806674836015,
      "grad_norm": 0.12958851461349183,
      "learning_rate": 0.0001144247361174141,
      "loss": 1.1484,
      "step": 6525
    },
    {
      "epoch": 0.470410149210697,
      "grad_norm": 0.11622847624663259,
      "learning_rate": 0.00011440163274579541,
      "loss": 1.1011,
      "step": 6526
    },
    {
      "epoch": 0.47048223167303393,
      "grad_norm": 0.20143274285403776,
      "learning_rate": 0.00011437852858915892,
      "loss": 1.1338,
      "step": 6527
    },
    {
      "epoch": 0.47055431413537085,
      "grad_norm": 0.1670178859756702,
      "learning_rate": 0.00011435542364876398,
      "loss": 1.0159,
      "step": 6528
    },
    {
      "epoch": 0.4706263965977078,
      "grad_norm": 0.177706124366429,
      "learning_rate": 0.00011433231792587003,
      "loss": 1.0565,
      "step": 6529
    },
    {
      "epoch": 0.4706984790600447,
      "grad_norm": 0.18461195910960657,
      "learning_rate": 0.00011430921142173652,
      "loss": 1.0121,
      "step": 6530
    },
    {
      "epoch": 0.4707705615223816,
      "grad_norm": 0.11695210646295924,
      "learning_rate": 0.000114286104137623,
      "loss": 1.0145,
      "step": 6531
    },
    {
      "epoch": 0.47084264398471853,
      "grad_norm": 0.12528556918110303,
      "learning_rate": 0.00011426299607478896,
      "loss": 1.0946,
      "step": 6532
    },
    {
      "epoch": 0.47091472644705545,
      "grad_norm": 0.3638908705949839,
      "learning_rate": 0.00011423988723449406,
      "loss": 1.0776,
      "step": 6533
    },
    {
      "epoch": 0.4709868089093923,
      "grad_norm": 0.1254169065355968,
      "learning_rate": 0.0001142167776179979,
      "loss": 1.0816,
      "step": 6534
    },
    {
      "epoch": 0.47105889137172924,
      "grad_norm": 0.1267418224119641,
      "learning_rate": 0.00011419366722656017,
      "loss": 1.0925,
      "step": 6535
    },
    {
      "epoch": 0.47113097383406616,
      "grad_norm": 0.11982335554495745,
      "learning_rate": 0.0001141705560614406,
      "loss": 1.0329,
      "step": 6536
    },
    {
      "epoch": 0.4712030562964031,
      "grad_norm": 0.14937168988265126,
      "learning_rate": 0.00011414744412389897,
      "loss": 0.9978,
      "step": 6537
    },
    {
      "epoch": 0.47127513875874,
      "grad_norm": 0.12219169779113775,
      "learning_rate": 0.00011412433141519504,
      "loss": 1.0314,
      "step": 6538
    },
    {
      "epoch": 0.4713472212210769,
      "grad_norm": 0.13545703626058594,
      "learning_rate": 0.0001141012179365887,
      "loss": 1.0723,
      "step": 6539
    },
    {
      "epoch": 0.47141930368341384,
      "grad_norm": 0.10685912566646993,
      "learning_rate": 0.00011407810368933984,
      "loss": 1.0282,
      "step": 6540
    },
    {
      "epoch": 0.47149138614575076,
      "grad_norm": 0.1353934302236081,
      "learning_rate": 0.00011405498867470839,
      "loss": 1.0933,
      "step": 6541
    },
    {
      "epoch": 0.4715634686080877,
      "grad_norm": 0.1108010734817167,
      "learning_rate": 0.0001140318728939543,
      "loss": 1.1015,
      "step": 6542
    },
    {
      "epoch": 0.47163555107042454,
      "grad_norm": 0.13605238524533916,
      "learning_rate": 0.00011400875634833758,
      "loss": 1.0817,
      "step": 6543
    },
    {
      "epoch": 0.47170763353276146,
      "grad_norm": 0.1283386941371555,
      "learning_rate": 0.00011398563903911835,
      "loss": 0.9666,
      "step": 6544
    },
    {
      "epoch": 0.4717797159950984,
      "grad_norm": 0.12362045738471353,
      "learning_rate": 0.00011396252096755668,
      "loss": 1.1338,
      "step": 6545
    },
    {
      "epoch": 0.4718517984574353,
      "grad_norm": 0.1418821968036805,
      "learning_rate": 0.00011393940213491267,
      "loss": 1.047,
      "step": 6546
    },
    {
      "epoch": 0.4719238809197722,
      "grad_norm": 0.11147259946376468,
      "learning_rate": 0.00011391628254244658,
      "loss": 1.044,
      "step": 6547
    },
    {
      "epoch": 0.47199596338210914,
      "grad_norm": 0.16015307182338087,
      "learning_rate": 0.00011389316219141854,
      "loss": 1.0285,
      "step": 6548
    },
    {
      "epoch": 0.47206804584444606,
      "grad_norm": 0.11092177744220007,
      "learning_rate": 0.00011387004108308892,
      "loss": 1.0807,
      "step": 6549
    },
    {
      "epoch": 0.472140128306783,
      "grad_norm": 0.11993817522491015,
      "learning_rate": 0.00011384691921871793,
      "loss": 1.0211,
      "step": 6550
    },
    {
      "epoch": 0.4722122107691199,
      "grad_norm": 0.13400056806973468,
      "learning_rate": 0.00011382379659956601,
      "loss": 0.9938,
      "step": 6551
    },
    {
      "epoch": 0.47228429323145676,
      "grad_norm": 0.12447769076117945,
      "learning_rate": 0.00011380067322689345,
      "loss": 1.0591,
      "step": 6552
    },
    {
      "epoch": 0.4723563756937937,
      "grad_norm": 0.12605492802259424,
      "learning_rate": 0.00011377754910196079,
      "loss": 1.0362,
      "step": 6553
    },
    {
      "epoch": 0.4724284581561306,
      "grad_norm": 0.27654926267666324,
      "learning_rate": 0.0001137544242260284,
      "loss": 0.9644,
      "step": 6554
    },
    {
      "epoch": 0.4725005406184675,
      "grad_norm": 0.16205096488342025,
      "learning_rate": 0.0001137312986003569,
      "loss": 1.0937,
      "step": 6555
    },
    {
      "epoch": 0.47257262308080444,
      "grad_norm": 0.13627860197172723,
      "learning_rate": 0.00011370817222620673,
      "loss": 1.2289,
      "step": 6556
    },
    {
      "epoch": 0.47264470554314136,
      "grad_norm": 0.15399622906960875,
      "learning_rate": 0.00011368504510483854,
      "loss": 1.0619,
      "step": 6557
    },
    {
      "epoch": 0.4727167880054783,
      "grad_norm": 0.11722605871655783,
      "learning_rate": 0.00011366191723751299,
      "loss": 1.0355,
      "step": 6558
    },
    {
      "epoch": 0.4727888704678152,
      "grad_norm": 0.14266707982652885,
      "learning_rate": 0.00011363878862549072,
      "loss": 1.1208,
      "step": 6559
    },
    {
      "epoch": 0.47286095293015207,
      "grad_norm": 0.13596161880276597,
      "learning_rate": 0.00011361565927003243,
      "loss": 0.9729,
      "step": 6560
    },
    {
      "epoch": 0.472933035392489,
      "grad_norm": 0.1267962665481778,
      "learning_rate": 0.00011359252917239893,
      "loss": 1.0632,
      "step": 6561
    },
    {
      "epoch": 0.4730051178548259,
      "grad_norm": 0.1294874966807739,
      "learning_rate": 0.00011356939833385097,
      "loss": 1.1365,
      "step": 6562
    },
    {
      "epoch": 0.4730772003171628,
      "grad_norm": 0.29686174529334913,
      "learning_rate": 0.00011354626675564942,
      "loss": 1.0351,
      "step": 6563
    },
    {
      "epoch": 0.47314928277949975,
      "grad_norm": 0.14823413450849543,
      "learning_rate": 0.00011352313443905511,
      "loss": 0.9771,
      "step": 6564
    },
    {
      "epoch": 0.47322136524183667,
      "grad_norm": 0.11198625526057943,
      "learning_rate": 0.00011350000138532902,
      "loss": 1.0377,
      "step": 6565
    },
    {
      "epoch": 0.4732934477041736,
      "grad_norm": 0.11754437478340438,
      "learning_rate": 0.00011347686759573206,
      "loss": 1.0521,
      "step": 6566
    },
    {
      "epoch": 0.4733655301665105,
      "grad_norm": 0.1078816749369349,
      "learning_rate": 0.00011345373307152529,
      "loss": 1.0679,
      "step": 6567
    },
    {
      "epoch": 0.4734376126288474,
      "grad_norm": 0.1397546082983219,
      "learning_rate": 0.00011343059781396965,
      "loss": 1.0605,
      "step": 6568
    },
    {
      "epoch": 0.4735096950911843,
      "grad_norm": 0.1357279648549026,
      "learning_rate": 0.00011340746182432633,
      "loss": 0.9988,
      "step": 6569
    },
    {
      "epoch": 0.4735817775535212,
      "grad_norm": 0.11284072434455834,
      "learning_rate": 0.00011338432510385635,
      "loss": 1.0108,
      "step": 6570
    },
    {
      "epoch": 0.47365386001585813,
      "grad_norm": 0.1252425363945672,
      "learning_rate": 0.00011336118765382095,
      "loss": 1.0361,
      "step": 6571
    },
    {
      "epoch": 0.47372594247819505,
      "grad_norm": 0.10543592400280927,
      "learning_rate": 0.00011333804947548125,
      "loss": 1.1476,
      "step": 6572
    },
    {
      "epoch": 0.47379802494053197,
      "grad_norm": 0.1448451428108527,
      "learning_rate": 0.00011331491057009858,
      "loss": 1.0514,
      "step": 6573
    },
    {
      "epoch": 0.4738701074028689,
      "grad_norm": 0.11324253625893618,
      "learning_rate": 0.00011329177093893411,
      "loss": 1.0938,
      "step": 6574
    },
    {
      "epoch": 0.4739421898652058,
      "grad_norm": 0.1005210852989326,
      "learning_rate": 0.00011326863058324926,
      "loss": 1.0623,
      "step": 6575
    },
    {
      "epoch": 0.47401427232754273,
      "grad_norm": 0.11577571032253486,
      "learning_rate": 0.00011324548950430529,
      "loss": 1.0552,
      "step": 6576
    },
    {
      "epoch": 0.47408635478987965,
      "grad_norm": 0.13182001280046066,
      "learning_rate": 0.00011322234770336368,
      "loss": 1.0236,
      "step": 6577
    },
    {
      "epoch": 0.4741584372522165,
      "grad_norm": 0.1364597944606267,
      "learning_rate": 0.00011319920518168584,
      "loss": 1.0896,
      "step": 6578
    },
    {
      "epoch": 0.47423051971455343,
      "grad_norm": 0.3879862803305502,
      "learning_rate": 0.0001131760619405332,
      "loss": 1.0152,
      "step": 6579
    },
    {
      "epoch": 0.47430260217689035,
      "grad_norm": 0.13286815045464762,
      "learning_rate": 0.00011315291798116732,
      "loss": 1.1036,
      "step": 6580
    },
    {
      "epoch": 0.4743746846392273,
      "grad_norm": 0.10685671035550327,
      "learning_rate": 0.00011312977330484975,
      "loss": 1.0084,
      "step": 6581
    },
    {
      "epoch": 0.4744467671015642,
      "grad_norm": 0.11799400578177745,
      "learning_rate": 0.00011310662791284205,
      "loss": 1.103,
      "step": 6582
    },
    {
      "epoch": 0.4745188495639011,
      "grad_norm": 0.11822536698525618,
      "learning_rate": 0.00011308348180640592,
      "loss": 0.958,
      "step": 6583
    },
    {
      "epoch": 0.47459093202623803,
      "grad_norm": 0.12167365023483129,
      "learning_rate": 0.00011306033498680292,
      "loss": 1.0692,
      "step": 6584
    },
    {
      "epoch": 0.47466301448857495,
      "grad_norm": 0.11484509420609411,
      "learning_rate": 0.00011303718745529485,
      "loss": 1.0688,
      "step": 6585
    },
    {
      "epoch": 0.4747350969509118,
      "grad_norm": 0.1396498269567533,
      "learning_rate": 0.00011301403921314345,
      "loss": 0.9865,
      "step": 6586
    },
    {
      "epoch": 0.47480717941324874,
      "grad_norm": 0.11967756916013643,
      "learning_rate": 0.00011299089026161047,
      "loss": 1.1231,
      "step": 6587
    },
    {
      "epoch": 0.47487926187558566,
      "grad_norm": 0.12164104602451521,
      "learning_rate": 0.00011296774060195776,
      "loss": 1.0759,
      "step": 6588
    },
    {
      "epoch": 0.4749513443379226,
      "grad_norm": 0.1417863779233146,
      "learning_rate": 0.00011294459023544717,
      "loss": 1.0406,
      "step": 6589
    },
    {
      "epoch": 0.4750234268002595,
      "grad_norm": 0.11403172539213,
      "learning_rate": 0.00011292143916334059,
      "loss": 0.9656,
      "step": 6590
    },
    {
      "epoch": 0.4750955092625964,
      "grad_norm": 0.14505933861936587,
      "learning_rate": 0.0001128982873869,
      "loss": 0.993,
      "step": 6591
    },
    {
      "epoch": 0.47516759172493334,
      "grad_norm": 0.15635745314136265,
      "learning_rate": 0.00011287513490738732,
      "loss": 1.0345,
      "step": 6592
    },
    {
      "epoch": 0.47523967418727026,
      "grad_norm": 0.12535169937392734,
      "learning_rate": 0.00011285198172606465,
      "loss": 1.082,
      "step": 6593
    },
    {
      "epoch": 0.4753117566496072,
      "grad_norm": 0.1340628123199284,
      "learning_rate": 0.00011282882784419398,
      "loss": 0.9609,
      "step": 6594
    },
    {
      "epoch": 0.47538383911194404,
      "grad_norm": 0.12039018124910991,
      "learning_rate": 0.00011280567326303742,
      "loss": 1.0309,
      "step": 6595
    },
    {
      "epoch": 0.47545592157428096,
      "grad_norm": 0.11970410237804521,
      "learning_rate": 0.00011278251798385708,
      "loss": 1.0643,
      "step": 6596
    },
    {
      "epoch": 0.4755280040366179,
      "grad_norm": 0.11892437731179567,
      "learning_rate": 0.00011275936200791522,
      "loss": 1.0993,
      "step": 6597
    },
    {
      "epoch": 0.4756000864989548,
      "grad_norm": 0.1113176056832975,
      "learning_rate": 0.00011273620533647391,
      "loss": 1.051,
      "step": 6598
    },
    {
      "epoch": 0.4756721689612917,
      "grad_norm": 0.10607981010140732,
      "learning_rate": 0.00011271304797079551,
      "loss": 1.0788,
      "step": 6599
    },
    {
      "epoch": 0.47574425142362864,
      "grad_norm": 0.12264211569421024,
      "learning_rate": 0.00011268988991214225,
      "loss": 1.0913,
      "step": 6600
    },
    {
      "epoch": 0.47581633388596556,
      "grad_norm": 0.11414129097216011,
      "learning_rate": 0.00011266673116177646,
      "loss": 1.0386,
      "step": 6601
    },
    {
      "epoch": 0.4758884163483025,
      "grad_norm": 1.7350259405959008,
      "learning_rate": 0.00011264357172096049,
      "loss": 1.0548,
      "step": 6602
    },
    {
      "epoch": 0.4759604988106394,
      "grad_norm": 0.11444872159050175,
      "learning_rate": 0.00011262041159095678,
      "loss": 1.0548,
      "step": 6603
    },
    {
      "epoch": 0.47603258127297626,
      "grad_norm": 0.12494903138786376,
      "learning_rate": 0.00011259725077302772,
      "loss": 1.1033,
      "step": 6604
    },
    {
      "epoch": 0.4761046637353132,
      "grad_norm": 0.10407593495040579,
      "learning_rate": 0.00011257408926843578,
      "loss": 1.0183,
      "step": 6605
    },
    {
      "epoch": 0.4761767461976501,
      "grad_norm": 0.1417233440422862,
      "learning_rate": 0.0001125509270784435,
      "loss": 1.0203,
      "step": 6606
    },
    {
      "epoch": 0.476248828659987,
      "grad_norm": 0.14010879104201127,
      "learning_rate": 0.00011252776420431338,
      "loss": 1.056,
      "step": 6607
    },
    {
      "epoch": 0.47632091112232394,
      "grad_norm": 0.11728277009796574,
      "learning_rate": 0.00011250460064730807,
      "loss": 1.0947,
      "step": 6608
    },
    {
      "epoch": 0.47639299358466086,
      "grad_norm": 0.1447552779918809,
      "learning_rate": 0.00011248143640869016,
      "loss": 1.0283,
      "step": 6609
    },
    {
      "epoch": 0.4764650760469978,
      "grad_norm": 0.35241192460646814,
      "learning_rate": 0.00011245827148972228,
      "loss": 1.0899,
      "step": 6610
    },
    {
      "epoch": 0.4765371585093347,
      "grad_norm": 0.11617840711422682,
      "learning_rate": 0.00011243510589166714,
      "loss": 1.0363,
      "step": 6611
    },
    {
      "epoch": 0.47660924097167157,
      "grad_norm": 0.1261126796559091,
      "learning_rate": 0.00011241193961578751,
      "loss": 1.1221,
      "step": 6612
    },
    {
      "epoch": 0.4766813234340085,
      "grad_norm": 0.1356394484127655,
      "learning_rate": 0.00011238877266334611,
      "loss": 1.0861,
      "step": 6613
    },
    {
      "epoch": 0.4767534058963454,
      "grad_norm": 0.10388444257688759,
      "learning_rate": 0.00011236560503560578,
      "loss": 1.0219,
      "step": 6614
    },
    {
      "epoch": 0.4768254883586823,
      "grad_norm": 0.17432086154023058,
      "learning_rate": 0.00011234243673382936,
      "loss": 1.0341,
      "step": 6615
    },
    {
      "epoch": 0.47689757082101925,
      "grad_norm": 0.3287432121789915,
      "learning_rate": 0.0001123192677592797,
      "loss": 1.0524,
      "step": 6616
    },
    {
      "epoch": 0.47696965328335617,
      "grad_norm": 0.12867944704414488,
      "learning_rate": 0.00011229609811321972,
      "loss": 1.0906,
      "step": 6617
    },
    {
      "epoch": 0.4770417357456931,
      "grad_norm": 0.12769736251551747,
      "learning_rate": 0.00011227292779691243,
      "loss": 1.062,
      "step": 6618
    },
    {
      "epoch": 0.47711381820803,
      "grad_norm": 0.10866011485078624,
      "learning_rate": 0.00011224975681162075,
      "loss": 1.0311,
      "step": 6619
    },
    {
      "epoch": 0.4771859006703669,
      "grad_norm": 0.11887961385697328,
      "learning_rate": 0.00011222658515860773,
      "loss": 1.0994,
      "step": 6620
    },
    {
      "epoch": 0.4772579831327038,
      "grad_norm": 0.12334921673677397,
      "learning_rate": 0.00011220341283913645,
      "loss": 1.0858,
      "step": 6621
    },
    {
      "epoch": 0.4773300655950407,
      "grad_norm": 0.16617256517780407,
      "learning_rate": 0.00011218023985447001,
      "loss": 0.9719,
      "step": 6622
    },
    {
      "epoch": 0.47740214805737763,
      "grad_norm": 0.3709201970100691,
      "learning_rate": 0.00011215706620587149,
      "loss": 1.0483,
      "step": 6623
    },
    {
      "epoch": 0.47747423051971455,
      "grad_norm": 0.7017249151355873,
      "learning_rate": 0.00011213389189460414,
      "loss": 0.9987,
      "step": 6624
    },
    {
      "epoch": 0.47754631298205147,
      "grad_norm": 0.19762979038825249,
      "learning_rate": 0.00011211071692193112,
      "loss": 1.1275,
      "step": 6625
    },
    {
      "epoch": 0.4776183954443884,
      "grad_norm": 0.12729718206128593,
      "learning_rate": 0.00011208754128911569,
      "loss": 1.0486,
      "step": 6626
    },
    {
      "epoch": 0.4776904779067253,
      "grad_norm": 0.14589522377044345,
      "learning_rate": 0.00011206436499742108,
      "loss": 1.1139,
      "step": 6627
    },
    {
      "epoch": 0.47776256036906223,
      "grad_norm": 0.12873059149362115,
      "learning_rate": 0.0001120411880481107,
      "loss": 1.1012,
      "step": 6628
    },
    {
      "epoch": 0.47783464283139915,
      "grad_norm": 0.27477002462630845,
      "learning_rate": 0.0001120180104424478,
      "loss": 1.0192,
      "step": 6629
    },
    {
      "epoch": 0.477906725293736,
      "grad_norm": 0.1172045598194109,
      "learning_rate": 0.00011199483218169587,
      "loss": 0.9982,
      "step": 6630
    },
    {
      "epoch": 0.47797880775607293,
      "grad_norm": 0.22811195419361494,
      "learning_rate": 0.0001119716532671182,
      "loss": 0.9882,
      "step": 6631
    },
    {
      "epoch": 0.47805089021840985,
      "grad_norm": 0.22247385147318394,
      "learning_rate": 0.0001119484736999784,
      "loss": 1.1015,
      "step": 6632
    },
    {
      "epoch": 0.47812297268074677,
      "grad_norm": 0.22987893744027896,
      "learning_rate": 0.00011192529348153986,
      "loss": 1.0609,
      "step": 6633
    },
    {
      "epoch": 0.4781950551430837,
      "grad_norm": 0.12648318692200405,
      "learning_rate": 0.00011190211261306613,
      "loss": 1.0631,
      "step": 6634
    },
    {
      "epoch": 0.4782671376054206,
      "grad_norm": 0.11466155102708281,
      "learning_rate": 0.0001118789310958208,
      "loss": 1.0158,
      "step": 6635
    },
    {
      "epoch": 0.47833922006775753,
      "grad_norm": 0.11312022156539325,
      "learning_rate": 0.00011185574893106745,
      "loss": 1.0885,
      "step": 6636
    },
    {
      "epoch": 0.47841130253009445,
      "grad_norm": 0.1300615522735442,
      "learning_rate": 0.00011183256612006971,
      "loss": 1.0979,
      "step": 6637
    },
    {
      "epoch": 0.4784833849924313,
      "grad_norm": 0.11250812048618332,
      "learning_rate": 0.00011180938266409126,
      "loss": 1.0307,
      "step": 6638
    },
    {
      "epoch": 0.47855546745476824,
      "grad_norm": 0.14031975235184632,
      "learning_rate": 0.0001117861985643958,
      "loss": 0.9996,
      "step": 6639
    },
    {
      "epoch": 0.47862754991710516,
      "grad_norm": 0.1451912664531764,
      "learning_rate": 0.00011176301382224709,
      "loss": 1.1249,
      "step": 6640
    },
    {
      "epoch": 0.4786996323794421,
      "grad_norm": 0.157603979648913,
      "learning_rate": 0.00011173982843890889,
      "loss": 1.0048,
      "step": 6641
    },
    {
      "epoch": 0.478771714841779,
      "grad_norm": 0.17314283324554847,
      "learning_rate": 0.00011171664241564498,
      "loss": 1.0345,
      "step": 6642
    },
    {
      "epoch": 0.4788437973041159,
      "grad_norm": 0.7168529499185773,
      "learning_rate": 0.00011169345575371927,
      "loss": 1.1436,
      "step": 6643
    },
    {
      "epoch": 0.47891587976645283,
      "grad_norm": 0.12874669391568808,
      "learning_rate": 0.00011167026845439563,
      "loss": 1.0692,
      "step": 6644
    },
    {
      "epoch": 0.47898796222878975,
      "grad_norm": 0.1256106883334417,
      "learning_rate": 0.00011164708051893791,
      "loss": 1.078,
      "step": 6645
    },
    {
      "epoch": 0.4790600446911267,
      "grad_norm": 0.19196846205949847,
      "learning_rate": 0.00011162389194861016,
      "loss": 1.0894,
      "step": 6646
    },
    {
      "epoch": 0.47913212715346354,
      "grad_norm": 0.16970825162746517,
      "learning_rate": 0.00011160070274467624,
      "loss": 1.0559,
      "step": 6647
    },
    {
      "epoch": 0.47920420961580046,
      "grad_norm": 0.13455677748200873,
      "learning_rate": 0.00011157751290840029,
      "loss": 1.0677,
      "step": 6648
    },
    {
      "epoch": 0.4792762920781374,
      "grad_norm": 0.12187889304075303,
      "learning_rate": 0.00011155432244104627,
      "loss": 1.0202,
      "step": 6649
    },
    {
      "epoch": 0.4793483745404743,
      "grad_norm": 0.1556307169049706,
      "learning_rate": 0.00011153113134387836,
      "loss": 1.0697,
      "step": 6650
    },
    {
      "epoch": 0.4794204570028112,
      "grad_norm": 0.1406676858545275,
      "learning_rate": 0.00011150793961816057,
      "loss": 1.037,
      "step": 6651
    },
    {
      "epoch": 0.47949253946514814,
      "grad_norm": 0.1564551416539053,
      "learning_rate": 0.00011148474726515716,
      "loss": 1.1083,
      "step": 6652
    },
    {
      "epoch": 0.47956462192748506,
      "grad_norm": 0.13532238425396853,
      "learning_rate": 0.00011146155428613225,
      "loss": 1.0189,
      "step": 6653
    },
    {
      "epoch": 0.479636704389822,
      "grad_norm": 0.6219352890847627,
      "learning_rate": 0.00011143836068235014,
      "loss": 1.0706,
      "step": 6654
    },
    {
      "epoch": 0.4797087868521589,
      "grad_norm": 0.11713997978318362,
      "learning_rate": 0.00011141516645507499,
      "loss": 1.0229,
      "step": 6655
    },
    {
      "epoch": 0.47978086931449576,
      "grad_norm": 0.12441368876050121,
      "learning_rate": 0.00011139197160557118,
      "loss": 1.046,
      "step": 6656
    },
    {
      "epoch": 0.4798529517768327,
      "grad_norm": 0.1390023179811271,
      "learning_rate": 0.00011136877613510299,
      "loss": 1.0169,
      "step": 6657
    },
    {
      "epoch": 0.4799250342391696,
      "grad_norm": 0.1701834068530347,
      "learning_rate": 0.00011134558004493478,
      "loss": 1.0999,
      "step": 6658
    },
    {
      "epoch": 0.4799971167015065,
      "grad_norm": 0.12548216881310834,
      "learning_rate": 0.00011132238333633098,
      "loss": 1.1144,
      "step": 6659
    },
    {
      "epoch": 0.48006919916384344,
      "grad_norm": 0.13436045097915666,
      "learning_rate": 0.00011129918601055598,
      "loss": 1.0927,
      "step": 6660
    },
    {
      "epoch": 0.48014128162618036,
      "grad_norm": 0.12563573232046282,
      "learning_rate": 0.00011127598806887427,
      "loss": 1.1069,
      "step": 6661
    },
    {
      "epoch": 0.4802133640885173,
      "grad_norm": 0.1142644959307002,
      "learning_rate": 0.00011125278951255032,
      "loss": 1.0278,
      "step": 6662
    },
    {
      "epoch": 0.4802854465508542,
      "grad_norm": 0.13732244143593242,
      "learning_rate": 0.00011122959034284868,
      "loss": 0.9923,
      "step": 6663
    },
    {
      "epoch": 0.48035752901319106,
      "grad_norm": 0.1269641287946793,
      "learning_rate": 0.00011120639056103388,
      "loss": 0.9986,
      "step": 6664
    },
    {
      "epoch": 0.480429611475528,
      "grad_norm": 0.14893750421923443,
      "learning_rate": 0.00011118319016837058,
      "loss": 1.0573,
      "step": 6665
    },
    {
      "epoch": 0.4805016939378649,
      "grad_norm": 0.1240860607697651,
      "learning_rate": 0.00011115998916612336,
      "loss": 1.0624,
      "step": 6666
    },
    {
      "epoch": 0.4805737764002018,
      "grad_norm": 0.11710264814159681,
      "learning_rate": 0.00011113678755555685,
      "loss": 1.0728,
      "step": 6667
    },
    {
      "epoch": 0.48064585886253874,
      "grad_norm": 0.11783487685329186,
      "learning_rate": 0.00011111358533793583,
      "loss": 1.0918,
      "step": 6668
    },
    {
      "epoch": 0.48071794132487566,
      "grad_norm": 0.142396059723051,
      "learning_rate": 0.00011109038251452495,
      "loss": 1.026,
      "step": 6669
    },
    {
      "epoch": 0.4807900237872126,
      "grad_norm": 0.1623856153771247,
      "learning_rate": 0.00011106717908658904,
      "loss": 1.0739,
      "step": 6670
    },
    {
      "epoch": 0.4808621062495495,
      "grad_norm": 0.19230615389970881,
      "learning_rate": 0.00011104397505539282,
      "loss": 1.0775,
      "step": 6671
    },
    {
      "epoch": 0.4809341887118864,
      "grad_norm": 0.12247309251091994,
      "learning_rate": 0.0001110207704222012,
      "loss": 1.0413,
      "step": 6672
    },
    {
      "epoch": 0.4810062711742233,
      "grad_norm": 0.243059930666288,
      "learning_rate": 0.00011099756518827896,
      "loss": 1.0788,
      "step": 6673
    },
    {
      "epoch": 0.4810783536365602,
      "grad_norm": 0.13114485487647623,
      "learning_rate": 0.00011097435935489107,
      "loss": 1.1065,
      "step": 6674
    },
    {
      "epoch": 0.4811504360988971,
      "grad_norm": 0.13495692514139507,
      "learning_rate": 0.00011095115292330236,
      "loss": 1.024,
      "step": 6675
    },
    {
      "epoch": 0.48122251856123405,
      "grad_norm": 0.11760510942445031,
      "learning_rate": 0.00011092794589477791,
      "loss": 1.0537,
      "step": 6676
    },
    {
      "epoch": 0.48129460102357097,
      "grad_norm": 0.13000808511302375,
      "learning_rate": 0.00011090473827058261,
      "loss": 1.07,
      "step": 6677
    },
    {
      "epoch": 0.4813666834859079,
      "grad_norm": 0.22506420094610302,
      "learning_rate": 0.0001108815300519815,
      "loss": 1.0135,
      "step": 6678
    },
    {
      "epoch": 0.4814387659482448,
      "grad_norm": 0.10763052642927726,
      "learning_rate": 0.00011085832124023966,
      "loss": 0.9987,
      "step": 6679
    },
    {
      "epoch": 0.4815108484105817,
      "grad_norm": 0.13499426910664,
      "learning_rate": 0.0001108351118366222,
      "loss": 1.0319,
      "step": 6680
    },
    {
      "epoch": 0.48158293087291865,
      "grad_norm": 0.14295809747764265,
      "learning_rate": 0.00011081190184239419,
      "loss": 1.0701,
      "step": 6681
    },
    {
      "epoch": 0.4816550133352555,
      "grad_norm": 0.14392648778686867,
      "learning_rate": 0.00011078869125882081,
      "loss": 1.0713,
      "step": 6682
    },
    {
      "epoch": 0.48172709579759243,
      "grad_norm": 0.22287141808802582,
      "learning_rate": 0.00011076548008716723,
      "loss": 1.0965,
      "step": 6683
    },
    {
      "epoch": 0.48179917825992935,
      "grad_norm": 0.13646437459060512,
      "learning_rate": 0.00011074226832869867,
      "loss": 1.0046,
      "step": 6684
    },
    {
      "epoch": 0.48187126072226627,
      "grad_norm": 0.10798428554468216,
      "learning_rate": 0.0001107190559846804,
      "loss": 1.052,
      "step": 6685
    },
    {
      "epoch": 0.4819433431846032,
      "grad_norm": 0.12326999019796606,
      "learning_rate": 0.00011069584305637768,
      "loss": 0.9835,
      "step": 6686
    },
    {
      "epoch": 0.4820154256469401,
      "grad_norm": 0.15834866153427224,
      "learning_rate": 0.00011067262954505582,
      "loss": 1.0732,
      "step": 6687
    },
    {
      "epoch": 0.48208750810927703,
      "grad_norm": 0.12218221467466181,
      "learning_rate": 0.00011064941545198018,
      "loss": 1.0313,
      "step": 6688
    },
    {
      "epoch": 0.48215959057161395,
      "grad_norm": 0.14384861530027016,
      "learning_rate": 0.0001106262007784161,
      "loss": 1.0983,
      "step": 6689
    },
    {
      "epoch": 0.4822316730339508,
      "grad_norm": 0.1899816315087201,
      "learning_rate": 0.00011060298552562905,
      "loss": 1.0909,
      "step": 6690
    },
    {
      "epoch": 0.48230375549628773,
      "grad_norm": 0.4302423609574336,
      "learning_rate": 0.00011057976969488439,
      "loss": 1.1529,
      "step": 6691
    },
    {
      "epoch": 0.48237583795862465,
      "grad_norm": 0.1279080833373373,
      "learning_rate": 0.00011055655328744767,
      "loss": 1.0181,
      "step": 6692
    },
    {
      "epoch": 0.4824479204209616,
      "grad_norm": 1.3188317241860412,
      "learning_rate": 0.00011053333630458434,
      "loss": 1.0761,
      "step": 6693
    },
    {
      "epoch": 0.4825200028832985,
      "grad_norm": 0.27540305405096793,
      "learning_rate": 0.00011051011874755995,
      "loss": 0.9986,
      "step": 6694
    },
    {
      "epoch": 0.4825920853456354,
      "grad_norm": 0.1318573433372885,
      "learning_rate": 0.00011048690061764005,
      "loss": 0.9939,
      "step": 6695
    },
    {
      "epoch": 0.48266416780797233,
      "grad_norm": 0.23842295173140407,
      "learning_rate": 0.00011046368191609025,
      "loss": 1.0292,
      "step": 6696
    },
    {
      "epoch": 0.48273625027030925,
      "grad_norm": 0.2732325264641157,
      "learning_rate": 0.00011044046264417618,
      "loss": 1.0399,
      "step": 6697
    },
    {
      "epoch": 0.4828083327326462,
      "grad_norm": 0.12079876265470868,
      "learning_rate": 0.0001104172428031635,
      "loss": 1.0605,
      "step": 6698
    },
    {
      "epoch": 0.48288041519498304,
      "grad_norm": 0.12424021875592288,
      "learning_rate": 0.00011039402239431791,
      "loss": 1.0483,
      "step": 6699
    },
    {
      "epoch": 0.48295249765731996,
      "grad_norm": 0.13621164544556205,
      "learning_rate": 0.00011037080141890505,
      "loss": 1.09,
      "step": 6700
    },
    {
      "epoch": 0.4830245801196569,
      "grad_norm": 0.12446869249716497,
      "learning_rate": 0.0001103475798781908,
      "loss": 0.9877,
      "step": 6701
    },
    {
      "epoch": 0.4830966625819938,
      "grad_norm": 0.17274180484190813,
      "learning_rate": 0.0001103243577734408,
      "loss": 1.0718,
      "step": 6702
    },
    {
      "epoch": 0.4831687450443307,
      "grad_norm": 0.10591864647594737,
      "learning_rate": 0.00011030113510592101,
      "loss": 1.0054,
      "step": 6703
    },
    {
      "epoch": 0.48324082750666764,
      "grad_norm": 0.12007664110506687,
      "learning_rate": 0.00011027791187689715,
      "loss": 1.0469,
      "step": 6704
    },
    {
      "epoch": 0.48331290996900456,
      "grad_norm": 0.1611975036551521,
      "learning_rate": 0.00011025468808763519,
      "loss": 1.1118,
      "step": 6705
    },
    {
      "epoch": 0.4833849924313415,
      "grad_norm": 0.13512027180293998,
      "learning_rate": 0.00011023146373940095,
      "loss": 1.0291,
      "step": 6706
    },
    {
      "epoch": 0.4834570748936784,
      "grad_norm": 0.14305713126146016,
      "learning_rate": 0.00011020823883346043,
      "loss": 1.09,
      "step": 6707
    },
    {
      "epoch": 0.48352915735601526,
      "grad_norm": 0.11662581173917991,
      "learning_rate": 0.00011018501337107954,
      "loss": 1.09,
      "step": 6708
    },
    {
      "epoch": 0.4836012398183522,
      "grad_norm": 0.14439597151014164,
      "learning_rate": 0.00011016178735352434,
      "loss": 1.0184,
      "step": 6709
    },
    {
      "epoch": 0.4836733222806891,
      "grad_norm": 0.1755770511205332,
      "learning_rate": 0.0001101385607820608,
      "loss": 1.0323,
      "step": 6710
    },
    {
      "epoch": 0.483745404743026,
      "grad_norm": 0.12867213419586673,
      "learning_rate": 0.00011011533365795498,
      "loss": 0.9989,
      "step": 6711
    },
    {
      "epoch": 0.48381748720536294,
      "grad_norm": 0.12404539516560363,
      "learning_rate": 0.00011009210598247302,
      "loss": 1.0426,
      "step": 6712
    },
    {
      "epoch": 0.48388956966769986,
      "grad_norm": 0.18473177151972317,
      "learning_rate": 0.00011006887775688095,
      "loss": 1.0217,
      "step": 6713
    },
    {
      "epoch": 0.4839616521300368,
      "grad_norm": 0.13110324973518084,
      "learning_rate": 0.00011004564898244498,
      "loss": 1.13,
      "step": 6714
    },
    {
      "epoch": 0.4840337345923737,
      "grad_norm": 0.1285049062777005,
      "learning_rate": 0.00011002241966043128,
      "loss": 1.053,
      "step": 6715
    },
    {
      "epoch": 0.48410581705471056,
      "grad_norm": 0.5049946054124398,
      "learning_rate": 0.00010999918979210604,
      "loss": 1.0853,
      "step": 6716
    },
    {
      "epoch": 0.4841778995170475,
      "grad_norm": 0.18846673089804078,
      "learning_rate": 0.0001099759593787355,
      "loss": 1.0388,
      "step": 6717
    },
    {
      "epoch": 0.4842499819793844,
      "grad_norm": 0.243919384163924,
      "learning_rate": 0.00010995272842158593,
      "loss": 1.139,
      "step": 6718
    },
    {
      "epoch": 0.4843220644417213,
      "grad_norm": 0.30776837547853675,
      "learning_rate": 0.00010992949692192362,
      "loss": 0.9873,
      "step": 6719
    },
    {
      "epoch": 0.48439414690405824,
      "grad_norm": 0.12059253739424243,
      "learning_rate": 0.0001099062648810149,
      "loss": 1.0949,
      "step": 6720
    },
    {
      "epoch": 0.48446622936639516,
      "grad_norm": 0.12255407015613046,
      "learning_rate": 0.00010988303230012613,
      "loss": 1.1056,
      "step": 6721
    },
    {
      "epoch": 0.4845383118287321,
      "grad_norm": 0.14894058532475749,
      "learning_rate": 0.00010985979918052364,
      "loss": 1.0867,
      "step": 6722
    },
    {
      "epoch": 0.484610394291069,
      "grad_norm": 0.1310341019221122,
      "learning_rate": 0.00010983656552347394,
      "loss": 1.0107,
      "step": 6723
    },
    {
      "epoch": 0.4846824767534059,
      "grad_norm": 0.14133537679585784,
      "learning_rate": 0.00010981333133024337,
      "loss": 1.0689,
      "step": 6724
    },
    {
      "epoch": 0.4847545592157428,
      "grad_norm": 0.11830879621440257,
      "learning_rate": 0.00010979009660209849,
      "loss": 1.014,
      "step": 6725
    },
    {
      "epoch": 0.4848266416780797,
      "grad_norm": 0.12139654593119084,
      "learning_rate": 0.00010976686134030574,
      "loss": 1.1019,
      "step": 6726
    },
    {
      "epoch": 0.4848987241404166,
      "grad_norm": 0.15620019181118752,
      "learning_rate": 0.0001097436255461317,
      "loss": 1.0778,
      "step": 6727
    },
    {
      "epoch": 0.48497080660275355,
      "grad_norm": 0.20980228982602203,
      "learning_rate": 0.00010972038922084283,
      "loss": 1.0713,
      "step": 6728
    },
    {
      "epoch": 0.48504288906509047,
      "grad_norm": 0.11445326492368024,
      "learning_rate": 0.00010969715236570586,
      "loss": 1.0504,
      "step": 6729
    },
    {
      "epoch": 0.4851149715274274,
      "grad_norm": 0.1323233071075236,
      "learning_rate": 0.00010967391498198729,
      "loss": 1.0451,
      "step": 6730
    },
    {
      "epoch": 0.4851870539897643,
      "grad_norm": 0.15753890602805842,
      "learning_rate": 0.00010965067707095382,
      "loss": 1.0333,
      "step": 6731
    },
    {
      "epoch": 0.4852591364521012,
      "grad_norm": 0.14165922587943944,
      "learning_rate": 0.0001096274386338721,
      "loss": 1.003,
      "step": 6732
    },
    {
      "epoch": 0.4853312189144381,
      "grad_norm": 0.14489421746727155,
      "learning_rate": 0.00010960419967200888,
      "loss": 1.0219,
      "step": 6733
    },
    {
      "epoch": 0.485403301376775,
      "grad_norm": 0.12993998823604236,
      "learning_rate": 0.00010958096018663083,
      "loss": 1.1166,
      "step": 6734
    },
    {
      "epoch": 0.48547538383911193,
      "grad_norm": 0.12963135099579143,
      "learning_rate": 0.00010955772017900474,
      "loss": 1.0377,
      "step": 6735
    },
    {
      "epoch": 0.48554746630144885,
      "grad_norm": 0.11783781806216191,
      "learning_rate": 0.00010953447965039737,
      "loss": 1.0193,
      "step": 6736
    },
    {
      "epoch": 0.48561954876378577,
      "grad_norm": 0.1509966590121591,
      "learning_rate": 0.0001095112386020756,
      "loss": 1.0843,
      "step": 6737
    },
    {
      "epoch": 0.4856916312261227,
      "grad_norm": 0.12248401761041983,
      "learning_rate": 0.00010948799703530621,
      "loss": 1.0722,
      "step": 6738
    },
    {
      "epoch": 0.4857637136884596,
      "grad_norm": 0.13169618989080717,
      "learning_rate": 0.0001094647549513561,
      "loss": 1.171,
      "step": 6739
    },
    {
      "epoch": 0.48583579615079653,
      "grad_norm": 0.12953031069495954,
      "learning_rate": 0.0001094415123514922,
      "loss": 1.1463,
      "step": 6740
    },
    {
      "epoch": 0.48590787861313345,
      "grad_norm": 0.6940871731461599,
      "learning_rate": 0.00010941826923698143,
      "loss": 1.1209,
      "step": 6741
    },
    {
      "epoch": 0.4859799610754703,
      "grad_norm": 0.1492219014902532,
      "learning_rate": 0.00010939502560909065,
      "loss": 1.0641,
      "step": 6742
    },
    {
      "epoch": 0.48605204353780723,
      "grad_norm": 0.13013929289711945,
      "learning_rate": 0.000109371781469087,
      "loss": 1.0211,
      "step": 6743
    },
    {
      "epoch": 0.48612412600014415,
      "grad_norm": 0.3820038737882887,
      "learning_rate": 0.00010934853681823736,
      "loss": 1.0265,
      "step": 6744
    },
    {
      "epoch": 0.48619620846248107,
      "grad_norm": 0.12213014101714313,
      "learning_rate": 0.00010932529165780887,
      "loss": 1.0611,
      "step": 6745
    },
    {
      "epoch": 0.486268290924818,
      "grad_norm": 0.1337451186363006,
      "learning_rate": 0.0001093020459890685,
      "loss": 1.0889,
      "step": 6746
    },
    {
      "epoch": 0.4863403733871549,
      "grad_norm": 0.1278723923183532,
      "learning_rate": 0.00010927879981328348,
      "loss": 1.0853,
      "step": 6747
    },
    {
      "epoch": 0.48641245584949183,
      "grad_norm": 0.12779258744880256,
      "learning_rate": 0.0001092555531317208,
      "loss": 1.0591,
      "step": 6748
    },
    {
      "epoch": 0.48648453831182875,
      "grad_norm": 0.21894688185033365,
      "learning_rate": 0.00010923230594564771,
      "loss": 1.0089,
      "step": 6749
    },
    {
      "epoch": 0.48655662077416567,
      "grad_norm": 0.5679426847880956,
      "learning_rate": 0.00010920905825633132,
      "loss": 1.0409,
      "step": 6750
    },
    {
      "epoch": 0.48662870323650254,
      "grad_norm": 0.5220822225331894,
      "learning_rate": 0.0001091858100650389,
      "loss": 1.0628,
      "step": 6751
    },
    {
      "epoch": 0.48670078569883946,
      "grad_norm": 0.17509438102005642,
      "learning_rate": 0.00010916256137303764,
      "loss": 1.057,
      "step": 6752
    },
    {
      "epoch": 0.4867728681611764,
      "grad_norm": 0.16517991294537335,
      "learning_rate": 0.00010913931218159481,
      "loss": 1.0489,
      "step": 6753
    },
    {
      "epoch": 0.4868449506235133,
      "grad_norm": 0.12350168476541391,
      "learning_rate": 0.00010911606249197773,
      "loss": 1.0148,
      "step": 6754
    },
    {
      "epoch": 0.4869170330858502,
      "grad_norm": 0.11132225528779008,
      "learning_rate": 0.00010909281230545365,
      "loss": 0.9795,
      "step": 6755
    },
    {
      "epoch": 0.48698911554818713,
      "grad_norm": 0.14221125014573965,
      "learning_rate": 0.00010906956162328996,
      "loss": 1.0142,
      "step": 6756
    },
    {
      "epoch": 0.48706119801052405,
      "grad_norm": 0.12348076265779852,
      "learning_rate": 0.00010904631044675404,
      "loss": 1.0593,
      "step": 6757
    },
    {
      "epoch": 0.487133280472861,
      "grad_norm": 0.11976898144491385,
      "learning_rate": 0.00010902305877711325,
      "loss": 1.0793,
      "step": 6758
    },
    {
      "epoch": 0.48720536293519784,
      "grad_norm": 0.19526229087182448,
      "learning_rate": 0.00010899980661563505,
      "loss": 1.0972,
      "step": 6759
    },
    {
      "epoch": 0.48727744539753476,
      "grad_norm": 0.15749804603553078,
      "learning_rate": 0.00010897655396358688,
      "loss": 0.9854,
      "step": 6760
    },
    {
      "epoch": 0.4873495278598717,
      "grad_norm": 0.3273277361739554,
      "learning_rate": 0.0001089533008222362,
      "loss": 1.0279,
      "step": 6761
    },
    {
      "epoch": 0.4874216103222086,
      "grad_norm": 0.1500951496431684,
      "learning_rate": 0.00010893004719285054,
      "loss": 1.0569,
      "step": 6762
    },
    {
      "epoch": 0.4874936927845455,
      "grad_norm": 0.17623157008020693,
      "learning_rate": 0.0001089067930766974,
      "loss": 1.0366,
      "step": 6763
    },
    {
      "epoch": 0.48756577524688244,
      "grad_norm": 0.15923056710851144,
      "learning_rate": 0.00010888353847504436,
      "loss": 1.0484,
      "step": 6764
    },
    {
      "epoch": 0.48763785770921936,
      "grad_norm": 0.11727313415915641,
      "learning_rate": 0.00010886028338915899,
      "loss": 1.0397,
      "step": 6765
    },
    {
      "epoch": 0.4877099401715563,
      "grad_norm": 0.1420760953279988,
      "learning_rate": 0.00010883702782030894,
      "loss": 1.0452,
      "step": 6766
    },
    {
      "epoch": 0.4877820226338932,
      "grad_norm": 0.11856313194780849,
      "learning_rate": 0.00010881377176976178,
      "loss": 1.0118,
      "step": 6767
    },
    {
      "epoch": 0.48785410509623006,
      "grad_norm": 0.12334280878568427,
      "learning_rate": 0.00010879051523878522,
      "loss": 0.9644,
      "step": 6768
    },
    {
      "epoch": 0.487926187558567,
      "grad_norm": 0.11708612804141612,
      "learning_rate": 0.00010876725822864695,
      "loss": 1.0835,
      "step": 6769
    },
    {
      "epoch": 0.4879982700209039,
      "grad_norm": 0.1292096755096176,
      "learning_rate": 0.00010874400074061466,
      "loss": 1.0232,
      "step": 6770
    },
    {
      "epoch": 0.4880703524832408,
      "grad_norm": 0.14815580392456904,
      "learning_rate": 0.0001087207427759561,
      "loss": 1.0019,
      "step": 6771
    },
    {
      "epoch": 0.48814243494557774,
      "grad_norm": 0.10839830062634727,
      "learning_rate": 0.00010869748433593905,
      "loss": 1.0307,
      "step": 6772
    },
    {
      "epoch": 0.48821451740791466,
      "grad_norm": 0.1597075379558655,
      "learning_rate": 0.0001086742254218313,
      "loss": 1.0175,
      "step": 6773
    },
    {
      "epoch": 0.4882865998702516,
      "grad_norm": 0.1259335244833207,
      "learning_rate": 0.00010865096603490066,
      "loss": 1.0366,
      "step": 6774
    },
    {
      "epoch": 0.4883586823325885,
      "grad_norm": 0.1280020116549302,
      "learning_rate": 0.00010862770617641495,
      "loss": 0.9784,
      "step": 6775
    },
    {
      "epoch": 0.4884307647949254,
      "grad_norm": 0.1308683101869091,
      "learning_rate": 0.00010860444584764211,
      "loss": 1.0197,
      "step": 6776
    },
    {
      "epoch": 0.4885028472572623,
      "grad_norm": 0.11057943657070418,
      "learning_rate": 0.00010858118504984995,
      "loss": 0.9963,
      "step": 6777
    },
    {
      "epoch": 0.4885749297195992,
      "grad_norm": 0.13678224696997632,
      "learning_rate": 0.0001085579237843065,
      "loss": 1.0608,
      "step": 6778
    },
    {
      "epoch": 0.4886470121819361,
      "grad_norm": 0.11579076681004438,
      "learning_rate": 0.00010853466205227955,
      "loss": 0.9887,
      "step": 6779
    },
    {
      "epoch": 0.48871909464427304,
      "grad_norm": 0.12956080224427058,
      "learning_rate": 0.00010851139985503724,
      "loss": 1.0656,
      "step": 6780
    },
    {
      "epoch": 0.48879117710660996,
      "grad_norm": 0.11520247556456348,
      "learning_rate": 0.00010848813719384745,
      "loss": 1.0514,
      "step": 6781
    },
    {
      "epoch": 0.4888632595689469,
      "grad_norm": 0.25910834017827306,
      "learning_rate": 0.0001084648740699783,
      "loss": 1.001,
      "step": 6782
    },
    {
      "epoch": 0.4889353420312838,
      "grad_norm": 0.15461800048613117,
      "learning_rate": 0.00010844161048469772,
      "loss": 1.0354,
      "step": 6783
    },
    {
      "epoch": 0.4890074244936207,
      "grad_norm": 0.11185212297862407,
      "learning_rate": 0.00010841834643927389,
      "loss": 1.0013,
      "step": 6784
    },
    {
      "epoch": 0.4890795069559576,
      "grad_norm": 0.20912161616844083,
      "learning_rate": 0.00010839508193497485,
      "loss": 1.1064,
      "step": 6785
    },
    {
      "epoch": 0.4891515894182945,
      "grad_norm": 0.15935363231586083,
      "learning_rate": 0.00010837181697306875,
      "loss": 1.0758,
      "step": 6786
    },
    {
      "epoch": 0.4892236718806314,
      "grad_norm": 0.1285701789995462,
      "learning_rate": 0.00010834855155482373,
      "loss": 1.0371,
      "step": 6787
    },
    {
      "epoch": 0.48929575434296835,
      "grad_norm": 0.11374609934053775,
      "learning_rate": 0.00010832528568150795,
      "loss": 1.0152,
      "step": 6788
    },
    {
      "epoch": 0.48936783680530527,
      "grad_norm": 0.15646292700683623,
      "learning_rate": 0.00010830201935438961,
      "loss": 1.0089,
      "step": 6789
    },
    {
      "epoch": 0.4894399192676422,
      "grad_norm": 0.12185622351980809,
      "learning_rate": 0.00010827875257473697,
      "loss": 0.988,
      "step": 6790
    },
    {
      "epoch": 0.4895120017299791,
      "grad_norm": 0.12298734709531652,
      "learning_rate": 0.00010825548534381824,
      "loss": 1.0872,
      "step": 6791
    },
    {
      "epoch": 0.489584084192316,
      "grad_norm": 0.11758992483516488,
      "learning_rate": 0.0001082322176629017,
      "loss": 1.0601,
      "step": 6792
    },
    {
      "epoch": 0.48965616665465295,
      "grad_norm": 0.14045760721681816,
      "learning_rate": 0.00010820894953325564,
      "loss": 1.0047,
      "step": 6793
    },
    {
      "epoch": 0.4897282491169898,
      "grad_norm": 0.12642043127174066,
      "learning_rate": 0.00010818568095614841,
      "loss": 1.1152,
      "step": 6794
    },
    {
      "epoch": 0.48980033157932673,
      "grad_norm": 0.1226484329005886,
      "learning_rate": 0.00010816241193284833,
      "loss": 1.0609,
      "step": 6795
    },
    {
      "epoch": 0.48987241404166365,
      "grad_norm": 0.1556448656859207,
      "learning_rate": 0.00010813914246462377,
      "loss": 1.151,
      "step": 6796
    },
    {
      "epoch": 0.48994449650400057,
      "grad_norm": 0.17890692394552324,
      "learning_rate": 0.00010811587255274313,
      "loss": 1.0568,
      "step": 6797
    },
    {
      "epoch": 0.4900165789663375,
      "grad_norm": 0.22592345663242405,
      "learning_rate": 0.00010809260219847485,
      "loss": 1.0463,
      "step": 6798
    },
    {
      "epoch": 0.4900886614286744,
      "grad_norm": 0.11986450733448371,
      "learning_rate": 0.00010806933140308733,
      "loss": 1.0546,
      "step": 6799
    },
    {
      "epoch": 0.49016074389101133,
      "grad_norm": 0.12326244336041951,
      "learning_rate": 0.00010804606016784908,
      "loss": 1.104,
      "step": 6800
    },
    {
      "epoch": 0.49023282635334825,
      "grad_norm": 0.165203215977399,
      "learning_rate": 0.00010802278849402854,
      "loss": 0.9842,
      "step": 6801
    },
    {
      "epoch": 0.49030490881568517,
      "grad_norm": 0.13112611341295458,
      "learning_rate": 0.00010799951638289427,
      "loss": 1.1355,
      "step": 6802
    },
    {
      "epoch": 0.49037699127802203,
      "grad_norm": 0.172671038271573,
      "learning_rate": 0.00010797624383571476,
      "loss": 1.017,
      "step": 6803
    },
    {
      "epoch": 0.49044907374035895,
      "grad_norm": 0.1572347011894616,
      "learning_rate": 0.00010795297085375866,
      "loss": 1.037,
      "step": 6804
    },
    {
      "epoch": 0.4905211562026959,
      "grad_norm": 0.16808203178144726,
      "learning_rate": 0.00010792969743829444,
      "loss": 1.061,
      "step": 6805
    },
    {
      "epoch": 0.4905932386650328,
      "grad_norm": 0.22666870575380219,
      "learning_rate": 0.00010790642359059085,
      "loss": 1.086,
      "step": 6806
    },
    {
      "epoch": 0.4906653211273697,
      "grad_norm": 0.2289104985202377,
      "learning_rate": 0.00010788314931191637,
      "loss": 1.0603,
      "step": 6807
    },
    {
      "epoch": 0.49073740358970663,
      "grad_norm": 0.12086915025918307,
      "learning_rate": 0.00010785987460353975,
      "loss": 1.0333,
      "step": 6808
    },
    {
      "epoch": 0.49080948605204355,
      "grad_norm": 0.1510084758064311,
      "learning_rate": 0.00010783659946672964,
      "loss": 0.9758,
      "step": 6809
    },
    {
      "epoch": 0.4908815685143805,
      "grad_norm": 0.1283606528604546,
      "learning_rate": 0.00010781332390275476,
      "loss": 1.0354,
      "step": 6810
    },
    {
      "epoch": 0.49095365097671734,
      "grad_norm": 0.11156262434114285,
      "learning_rate": 0.00010779004791288383,
      "loss": 1.02,
      "step": 6811
    },
    {
      "epoch": 0.49102573343905426,
      "grad_norm": 0.6397912953990473,
      "learning_rate": 0.00010776677149838559,
      "loss": 1.0416,
      "step": 6812
    },
    {
      "epoch": 0.4910978159013912,
      "grad_norm": 0.11610576971339956,
      "learning_rate": 0.00010774349466052882,
      "loss": 1.0244,
      "step": 6813
    },
    {
      "epoch": 0.4911698983637281,
      "grad_norm": 0.12148767034007488,
      "learning_rate": 0.00010772021740058231,
      "loss": 1.0656,
      "step": 6814
    },
    {
      "epoch": 0.491241980826065,
      "grad_norm": 0.12098668037664172,
      "learning_rate": 0.0001076969397198149,
      "loss": 1.0118,
      "step": 6815
    },
    {
      "epoch": 0.49131406328840194,
      "grad_norm": 0.11245702707622819,
      "learning_rate": 0.00010767366161949542,
      "loss": 1.0349,
      "step": 6816
    },
    {
      "epoch": 0.49138614575073886,
      "grad_norm": 0.11462098363688637,
      "learning_rate": 0.0001076503831008927,
      "loss": 1.0618,
      "step": 6817
    },
    {
      "epoch": 0.4914582282130758,
      "grad_norm": 0.43609302837971337,
      "learning_rate": 0.00010762710416527571,
      "loss": 1.0816,
      "step": 6818
    },
    {
      "epoch": 0.4915303106754127,
      "grad_norm": 0.15746615470077438,
      "learning_rate": 0.00010760382481391325,
      "loss": 0.987,
      "step": 6819
    },
    {
      "epoch": 0.49160239313774956,
      "grad_norm": 0.18922722378772405,
      "learning_rate": 0.00010758054504807435,
      "loss": 1.0873,
      "step": 6820
    },
    {
      "epoch": 0.4916744756000865,
      "grad_norm": 0.10349684572392649,
      "learning_rate": 0.0001075572648690279,
      "loss": 0.9705,
      "step": 6821
    },
    {
      "epoch": 0.4917465580624234,
      "grad_norm": 0.11324807761860008,
      "learning_rate": 0.00010753398427804293,
      "loss": 1.0464,
      "step": 6822
    },
    {
      "epoch": 0.4918186405247603,
      "grad_norm": 0.12880091772110294,
      "learning_rate": 0.0001075107032763884,
      "loss": 1.0728,
      "step": 6823
    },
    {
      "epoch": 0.49189072298709724,
      "grad_norm": 0.11258823267171032,
      "learning_rate": 0.00010748742186533338,
      "loss": 1.0019,
      "step": 6824
    },
    {
      "epoch": 0.49196280544943416,
      "grad_norm": 0.15527885614876383,
      "learning_rate": 0.00010746414004614683,
      "loss": 1.1972,
      "step": 6825
    },
    {
      "epoch": 0.4920348879117711,
      "grad_norm": 0.14779021281344856,
      "learning_rate": 0.00010744085782009792,
      "loss": 0.9508,
      "step": 6826
    },
    {
      "epoch": 0.492106970374108,
      "grad_norm": 0.1282180644410377,
      "learning_rate": 0.00010741757518845565,
      "loss": 1.0496,
      "step": 6827
    },
    {
      "epoch": 0.4921790528364449,
      "grad_norm": 0.17855783519919824,
      "learning_rate": 0.00010739429215248922,
      "loss": 1.0062,
      "step": 6828
    },
    {
      "epoch": 0.4922511352987818,
      "grad_norm": 0.2842768738667225,
      "learning_rate": 0.0001073710087134677,
      "loss": 1.0527,
      "step": 6829
    },
    {
      "epoch": 0.4923232177611187,
      "grad_norm": 0.11974546723835625,
      "learning_rate": 0.00010734772487266026,
      "loss": 1.1295,
      "step": 6830
    },
    {
      "epoch": 0.4923953002234556,
      "grad_norm": 0.16654363054143523,
      "learning_rate": 0.00010732444063133607,
      "loss": 1.0352,
      "step": 6831
    },
    {
      "epoch": 0.49246738268579254,
      "grad_norm": 0.16990888049070604,
      "learning_rate": 0.00010730115599076434,
      "loss": 1.0104,
      "step": 6832
    },
    {
      "epoch": 0.49253946514812946,
      "grad_norm": 0.10772781459672821,
      "learning_rate": 0.00010727787095221429,
      "loss": 1.0451,
      "step": 6833
    },
    {
      "epoch": 0.4926115476104664,
      "grad_norm": 0.1309317383603401,
      "learning_rate": 0.00010725458551695515,
      "loss": 0.9873,
      "step": 6834
    },
    {
      "epoch": 0.4926836300728033,
      "grad_norm": 0.12747248818097254,
      "learning_rate": 0.00010723129968625623,
      "loss": 1.0287,
      "step": 6835
    },
    {
      "epoch": 0.4927557125351402,
      "grad_norm": 0.11958289584856009,
      "learning_rate": 0.00010720801346138676,
      "loss": 1.0071,
      "step": 6836
    },
    {
      "epoch": 0.4928277949974771,
      "grad_norm": 0.14082275822715098,
      "learning_rate": 0.00010718472684361609,
      "loss": 1.0284,
      "step": 6837
    },
    {
      "epoch": 0.492899877459814,
      "grad_norm": 0.15502251083563087,
      "learning_rate": 0.00010716143983421352,
      "loss": 1.0834,
      "step": 6838
    },
    {
      "epoch": 0.4929719599221509,
      "grad_norm": 0.1711709427730406,
      "learning_rate": 0.00010713815243444841,
      "loss": 0.9964,
      "step": 6839
    },
    {
      "epoch": 0.49304404238448785,
      "grad_norm": 0.1302807391350465,
      "learning_rate": 0.00010711486464559013,
      "loss": 1.1164,
      "step": 6840
    },
    {
      "epoch": 0.49311612484682477,
      "grad_norm": 0.17191235551685824,
      "learning_rate": 0.00010709157646890809,
      "loss": 1.0522,
      "step": 6841
    },
    {
      "epoch": 0.4931882073091617,
      "grad_norm": 0.1257076482549578,
      "learning_rate": 0.00010706828790567167,
      "loss": 1.0476,
      "step": 6842
    },
    {
      "epoch": 0.4932602897714986,
      "grad_norm": 0.13166378840271023,
      "learning_rate": 0.00010704499895715032,
      "loss": 1.1039,
      "step": 6843
    },
    {
      "epoch": 0.4933323722338355,
      "grad_norm": 0.12063352557452522,
      "learning_rate": 0.00010702170962461354,
      "loss": 0.981,
      "step": 6844
    },
    {
      "epoch": 0.49340445469617245,
      "grad_norm": 0.14742085037681862,
      "learning_rate": 0.00010699841990933073,
      "loss": 1.0994,
      "step": 6845
    },
    {
      "epoch": 0.4934765371585093,
      "grad_norm": 0.12040088733285664,
      "learning_rate": 0.00010697512981257143,
      "loss": 1.0732,
      "step": 6846
    },
    {
      "epoch": 0.49354861962084623,
      "grad_norm": 0.13003883136479913,
      "learning_rate": 0.00010695183933560515,
      "loss": 1.1303,
      "step": 6847
    },
    {
      "epoch": 0.49362070208318315,
      "grad_norm": 0.19350128634327698,
      "learning_rate": 0.00010692854847970147,
      "loss": 1.0452,
      "step": 6848
    },
    {
      "epoch": 0.49369278454552007,
      "grad_norm": 0.44134309153825474,
      "learning_rate": 0.00010690525724612987,
      "loss": 1.1147,
      "step": 6849
    },
    {
      "epoch": 0.493764867007857,
      "grad_norm": 0.1491485368717767,
      "learning_rate": 0.00010688196563616,
      "loss": 0.961,
      "step": 6850
    },
    {
      "epoch": 0.4938369494701939,
      "grad_norm": 0.1259928257477409,
      "learning_rate": 0.00010685867365106143,
      "loss": 0.941,
      "step": 6851
    },
    {
      "epoch": 0.49390903193253083,
      "grad_norm": 0.18294867970827317,
      "learning_rate": 0.00010683538129210377,
      "loss": 1.0333,
      "step": 6852
    },
    {
      "epoch": 0.49398111439486775,
      "grad_norm": 0.1445043273391062,
      "learning_rate": 0.00010681208856055671,
      "loss": 1.1002,
      "step": 6853
    },
    {
      "epoch": 0.49405319685720467,
      "grad_norm": 0.6294728734345066,
      "learning_rate": 0.00010678879545768985,
      "loss": 1.0147,
      "step": 6854
    },
    {
      "epoch": 0.49412527931954153,
      "grad_norm": 0.1218081768703105,
      "learning_rate": 0.00010676550198477293,
      "loss": 1.0546,
      "step": 6855
    },
    {
      "epoch": 0.49419736178187845,
      "grad_norm": 0.1194454466555216,
      "learning_rate": 0.00010674220814307559,
      "loss": 1.1009,
      "step": 6856
    },
    {
      "epoch": 0.4942694442442154,
      "grad_norm": 0.13098606914943575,
      "learning_rate": 0.00010671891393386764,
      "loss": 1.0679,
      "step": 6857
    },
    {
      "epoch": 0.4943415267065523,
      "grad_norm": 0.11977577211924502,
      "learning_rate": 0.00010669561935841871,
      "loss": 1.1118,
      "step": 6858
    },
    {
      "epoch": 0.4944136091688892,
      "grad_norm": 0.11394074914244809,
      "learning_rate": 0.00010667232441799867,
      "loss": 0.9857,
      "step": 6859
    },
    {
      "epoch": 0.49448569163122613,
      "grad_norm": 0.12913488599594536,
      "learning_rate": 0.00010664902911387722,
      "loss": 1.0973,
      "step": 6860
    },
    {
      "epoch": 0.49455777409356305,
      "grad_norm": 0.1571517165244327,
      "learning_rate": 0.00010662573344732425,
      "loss": 1.0332,
      "step": 6861
    },
    {
      "epoch": 0.49462985655589997,
      "grad_norm": 0.142896676656198,
      "learning_rate": 0.00010660243741960952,
      "loss": 1.0686,
      "step": 6862
    },
    {
      "epoch": 0.49470193901823684,
      "grad_norm": 0.14636050770909179,
      "learning_rate": 0.00010657914103200286,
      "loss": 1.0341,
      "step": 6863
    },
    {
      "epoch": 0.49477402148057376,
      "grad_norm": 0.1199617191973638,
      "learning_rate": 0.00010655584428577415,
      "loss": 0.9747,
      "step": 6864
    },
    {
      "epoch": 0.4948461039429107,
      "grad_norm": 0.11582198463953076,
      "learning_rate": 0.0001065325471821933,
      "loss": 1.0966,
      "step": 6865
    },
    {
      "epoch": 0.4949181864052476,
      "grad_norm": 0.8887733170933496,
      "learning_rate": 0.00010650924972253016,
      "loss": 1.1231,
      "step": 6866
    },
    {
      "epoch": 0.4949902688675845,
      "grad_norm": 0.12756223488309476,
      "learning_rate": 0.00010648595190805471,
      "loss": 1.0639,
      "step": 6867
    },
    {
      "epoch": 0.49506235132992144,
      "grad_norm": 0.1260954825752089,
      "learning_rate": 0.00010646265374003683,
      "loss": 0.9876,
      "step": 6868
    },
    {
      "epoch": 0.49513443379225835,
      "grad_norm": 0.2877108047681528,
      "learning_rate": 0.00010643935521974651,
      "loss": 1.009,
      "step": 6869
    },
    {
      "epoch": 0.4952065162545953,
      "grad_norm": 1.3122233051008612,
      "learning_rate": 0.00010641605634845373,
      "loss": 1.0551,
      "step": 6870
    },
    {
      "epoch": 0.4952785987169322,
      "grad_norm": 0.2009563797934317,
      "learning_rate": 0.00010639275712742847,
      "loss": 1.0065,
      "step": 6871
    },
    {
      "epoch": 0.49535068117926906,
      "grad_norm": 0.14303220461624172,
      "learning_rate": 0.00010636945755794074,
      "loss": 1.0744,
      "step": 6872
    },
    {
      "epoch": 0.495422763641606,
      "grad_norm": 0.10422826239081352,
      "learning_rate": 0.00010634615764126063,
      "loss": 1.092,
      "step": 6873
    },
    {
      "epoch": 0.4954948461039429,
      "grad_norm": 0.11804779866863907,
      "learning_rate": 0.00010632285737865813,
      "loss": 1.0178,
      "step": 6874
    },
    {
      "epoch": 0.4955669285662798,
      "grad_norm": 0.11632467307883157,
      "learning_rate": 0.00010629955677140335,
      "loss": 1.0365,
      "step": 6875
    },
    {
      "epoch": 0.49563901102861674,
      "grad_norm": 0.11761978099857118,
      "learning_rate": 0.00010627625582076634,
      "loss": 1.0694,
      "step": 6876
    },
    {
      "epoch": 0.49571109349095366,
      "grad_norm": 0.20684596966994584,
      "learning_rate": 0.0001062529545280173,
      "loss": 1.07,
      "step": 6877
    },
    {
      "epoch": 0.4957831759532906,
      "grad_norm": 0.10868997923504131,
      "learning_rate": 0.00010622965289442622,
      "loss": 1.0384,
      "step": 6878
    },
    {
      "epoch": 0.4958552584156275,
      "grad_norm": 0.14499547888361194,
      "learning_rate": 0.00010620635092126338,
      "loss": 1.0674,
      "step": 6879
    },
    {
      "epoch": 0.4959273408779644,
      "grad_norm": 0.12660168237497596,
      "learning_rate": 0.00010618304860979884,
      "loss": 1.0136,
      "step": 6880
    },
    {
      "epoch": 0.4959994233403013,
      "grad_norm": 0.14464347075923717,
      "learning_rate": 0.0001061597459613029,
      "loss": 0.9746,
      "step": 6881
    },
    {
      "epoch": 0.4960715058026382,
      "grad_norm": 0.11945118178492015,
      "learning_rate": 0.00010613644297704563,
      "loss": 1.0568,
      "step": 6882
    },
    {
      "epoch": 0.4961435882649751,
      "grad_norm": 0.1179679452083718,
      "learning_rate": 0.00010611313965829738,
      "loss": 0.9864,
      "step": 6883
    },
    {
      "epoch": 0.49621567072731204,
      "grad_norm": 0.14590654030337494,
      "learning_rate": 0.00010608983600632831,
      "loss": 1.0921,
      "step": 6884
    },
    {
      "epoch": 0.49628775318964896,
      "grad_norm": 0.11976545542935528,
      "learning_rate": 0.00010606653202240871,
      "loss": 0.9956,
      "step": 6885
    },
    {
      "epoch": 0.4963598356519859,
      "grad_norm": 0.1358532504324996,
      "learning_rate": 0.00010604322770780881,
      "loss": 1.1084,
      "step": 6886
    },
    {
      "epoch": 0.4964319181143228,
      "grad_norm": 0.1304325887524262,
      "learning_rate": 0.00010601992306379896,
      "loss": 1.1254,
      "step": 6887
    },
    {
      "epoch": 0.4965040005766597,
      "grad_norm": 0.1456844197215752,
      "learning_rate": 0.00010599661809164943,
      "loss": 1.0901,
      "step": 6888
    },
    {
      "epoch": 0.4965760830389966,
      "grad_norm": 0.12347359075190295,
      "learning_rate": 0.00010597331279263057,
      "loss": 1.1024,
      "step": 6889
    },
    {
      "epoch": 0.4966481655013335,
      "grad_norm": 0.15783430585781832,
      "learning_rate": 0.00010595000716801274,
      "loss": 1.0416,
      "step": 6890
    },
    {
      "epoch": 0.4967202479636704,
      "grad_norm": 0.1265781187571333,
      "learning_rate": 0.00010592670121906625,
      "loss": 1.0317,
      "step": 6891
    },
    {
      "epoch": 0.49679233042600734,
      "grad_norm": 0.12270061705956338,
      "learning_rate": 0.00010590339494706157,
      "loss": 0.9216,
      "step": 6892
    },
    {
      "epoch": 0.49686441288834426,
      "grad_norm": 0.14045769764842295,
      "learning_rate": 0.00010588008835326902,
      "loss": 1.0987,
      "step": 6893
    },
    {
      "epoch": 0.4969364953506812,
      "grad_norm": 1.22603580483449,
      "learning_rate": 0.00010585678143895906,
      "loss": 0.9725,
      "step": 6894
    },
    {
      "epoch": 0.4970085778130181,
      "grad_norm": 0.11841756892727147,
      "learning_rate": 0.00010583347420540213,
      "loss": 1.0552,
      "step": 6895
    },
    {
      "epoch": 0.497080660275355,
      "grad_norm": 0.11894625380107249,
      "learning_rate": 0.00010581016665386862,
      "loss": 0.9909,
      "step": 6896
    },
    {
      "epoch": 0.49715274273769194,
      "grad_norm": 0.25121299720549184,
      "learning_rate": 0.00010578685878562912,
      "loss": 1.0639,
      "step": 6897
    },
    {
      "epoch": 0.4972248252000288,
      "grad_norm": 0.1189444707131491,
      "learning_rate": 0.00010576355060195397,
      "loss": 1.097,
      "step": 6898
    },
    {
      "epoch": 0.49729690766236573,
      "grad_norm": 0.11044482175339009,
      "learning_rate": 0.00010574024210411382,
      "loss": 1.0855,
      "step": 6899
    },
    {
      "epoch": 0.49736899012470265,
      "grad_norm": 0.15824854074442987,
      "learning_rate": 0.00010571693329337906,
      "loss": 1.1274,
      "step": 6900
    },
    {
      "epoch": 0.49744107258703957,
      "grad_norm": 0.12017180125654903,
      "learning_rate": 0.00010569362417102035,
      "loss": 1.0369,
      "step": 6901
    },
    {
      "epoch": 0.4975131550493765,
      "grad_norm": 0.12485909008261516,
      "learning_rate": 0.00010567031473830813,
      "loss": 1.0595,
      "step": 6902
    },
    {
      "epoch": 0.4975852375117134,
      "grad_norm": 0.12874602948894073,
      "learning_rate": 0.00010564700499651308,
      "loss": 0.9914,
      "step": 6903
    },
    {
      "epoch": 0.4976573199740503,
      "grad_norm": 0.12911828491058144,
      "learning_rate": 0.00010562369494690569,
      "loss": 1.036,
      "step": 6904
    },
    {
      "epoch": 0.49772940243638725,
      "grad_norm": 0.12315914521952477,
      "learning_rate": 0.00010560038459075665,
      "loss": 0.9719,
      "step": 6905
    },
    {
      "epoch": 0.49780148489872417,
      "grad_norm": 0.13599436911396823,
      "learning_rate": 0.00010557707392933654,
      "loss": 1.0521,
      "step": 6906
    },
    {
      "epoch": 0.49787356736106103,
      "grad_norm": 0.1464692732508982,
      "learning_rate": 0.00010555376296391599,
      "loss": 1.0694,
      "step": 6907
    },
    {
      "epoch": 0.49794564982339795,
      "grad_norm": 1.3893201222103633,
      "learning_rate": 0.0001055304516957657,
      "loss": 1.1062,
      "step": 6908
    },
    {
      "epoch": 0.49801773228573487,
      "grad_norm": 0.11039775102137606,
      "learning_rate": 0.00010550714012615628,
      "loss": 1.1058,
      "step": 6909
    },
    {
      "epoch": 0.4980898147480718,
      "grad_norm": 0.10936356591641924,
      "learning_rate": 0.00010548382825635847,
      "loss": 1.0832,
      "step": 6910
    },
    {
      "epoch": 0.4981618972104087,
      "grad_norm": 0.1290961483393107,
      "learning_rate": 0.00010546051608764298,
      "loss": 1.1315,
      "step": 6911
    },
    {
      "epoch": 0.49823397967274563,
      "grad_norm": 0.15430391582143801,
      "learning_rate": 0.0001054372036212805,
      "loss": 1.0857,
      "step": 6912
    },
    {
      "epoch": 0.49830606213508255,
      "grad_norm": 0.13468567268090587,
      "learning_rate": 0.00010541389085854176,
      "loss": 1.0501,
      "step": 6913
    },
    {
      "epoch": 0.49837814459741947,
      "grad_norm": 0.11693390869585998,
      "learning_rate": 0.00010539057780069757,
      "loss": 1.0114,
      "step": 6914
    },
    {
      "epoch": 0.49845022705975633,
      "grad_norm": 0.13246355310856137,
      "learning_rate": 0.00010536726444901863,
      "loss": 1.1246,
      "step": 6915
    },
    {
      "epoch": 0.49852230952209325,
      "grad_norm": 0.19335007458815986,
      "learning_rate": 0.00010534395080477577,
      "loss": 0.9956,
      "step": 6916
    },
    {
      "epoch": 0.4985943919844302,
      "grad_norm": 0.1299884544302219,
      "learning_rate": 0.00010532063686923979,
      "loss": 1.0178,
      "step": 6917
    },
    {
      "epoch": 0.4986664744467671,
      "grad_norm": 0.12498434964333612,
      "learning_rate": 0.00010529732264368151,
      "loss": 1.0208,
      "step": 6918
    },
    {
      "epoch": 0.498738556909104,
      "grad_norm": 0.16635011876685468,
      "learning_rate": 0.00010527400812937174,
      "loss": 1.0291,
      "step": 6919
    },
    {
      "epoch": 0.49881063937144093,
      "grad_norm": 0.16952491947882195,
      "learning_rate": 0.00010525069332758135,
      "loss": 1.075,
      "step": 6920
    },
    {
      "epoch": 0.49888272183377785,
      "grad_norm": 1.4369733985703665,
      "learning_rate": 0.0001052273782395812,
      "loss": 1.0257,
      "step": 6921
    },
    {
      "epoch": 0.4989548042961148,
      "grad_norm": 0.12188248435194671,
      "learning_rate": 0.00010520406286664215,
      "loss": 1.0417,
      "step": 6922
    },
    {
      "epoch": 0.4990268867584517,
      "grad_norm": 0.1677982619072069,
      "learning_rate": 0.00010518074721003514,
      "loss": 1.1455,
      "step": 6923
    },
    {
      "epoch": 0.49909896922078856,
      "grad_norm": 0.1413345861093121,
      "learning_rate": 0.00010515743127103106,
      "loss": 1.0057,
      "step": 6924
    },
    {
      "epoch": 0.4991710516831255,
      "grad_norm": 0.14783667820935203,
      "learning_rate": 0.00010513411505090083,
      "loss": 1.0707,
      "step": 6925
    },
    {
      "epoch": 0.4992431341454624,
      "grad_norm": 0.11632598636678906,
      "learning_rate": 0.00010511079855091539,
      "loss": 1.0635,
      "step": 6926
    },
    {
      "epoch": 0.4993152166077993,
      "grad_norm": 0.15383291869689386,
      "learning_rate": 0.00010508748177234574,
      "loss": 1.1136,
      "step": 6927
    },
    {
      "epoch": 0.49938729907013624,
      "grad_norm": 0.12157997599851879,
      "learning_rate": 0.00010506416471646278,
      "loss": 1.0031,
      "step": 6928
    },
    {
      "epoch": 0.49945938153247316,
      "grad_norm": 0.17962085633168956,
      "learning_rate": 0.00010504084738453755,
      "loss": 1.041,
      "step": 6929
    },
    {
      "epoch": 0.4995314639948101,
      "grad_norm": 0.10961226334583586,
      "learning_rate": 0.00010501752977784105,
      "loss": 1.0653,
      "step": 6930
    },
    {
      "epoch": 0.499603546457147,
      "grad_norm": 0.22569413552228296,
      "learning_rate": 0.00010499421189764427,
      "loss": 1.0723,
      "step": 6931
    },
    {
      "epoch": 0.4996756289194839,
      "grad_norm": 0.1292014863656501,
      "learning_rate": 0.00010497089374521829,
      "loss": 0.9897,
      "step": 6932
    },
    {
      "epoch": 0.4997477113818208,
      "grad_norm": 0.36364261344240595,
      "learning_rate": 0.0001049475753218341,
      "loss": 1.0399,
      "step": 6933
    },
    {
      "epoch": 0.4998197938441577,
      "grad_norm": 0.14448513874180868,
      "learning_rate": 0.00010492425662876282,
      "loss": 1.0881,
      "step": 6934
    },
    {
      "epoch": 0.4998918763064946,
      "grad_norm": 0.13318014113214183,
      "learning_rate": 0.00010490093766727546,
      "loss": 0.9642,
      "step": 6935
    },
    {
      "epoch": 0.49996395876883154,
      "grad_norm": 0.5567138337911761,
      "learning_rate": 0.00010487761843864321,
      "loss": 0.9824,
      "step": 6936
    },
    {
      "epoch": 0.5000360412311684,
      "grad_norm": 0.13304998381204605,
      "learning_rate": 0.00010485429894413707,
      "loss": 1.0712,
      "step": 6937
    },
    {
      "epoch": 0.5001081236935053,
      "grad_norm": 0.4110358697267302,
      "learning_rate": 0.00010483097918502824,
      "loss": 1.1063,
      "step": 6938
    },
    {
      "epoch": 0.5001802061558422,
      "grad_norm": 0.185805324505842,
      "learning_rate": 0.00010480765916258781,
      "loss": 1.0202,
      "step": 6939
    },
    {
      "epoch": 0.5002522886181792,
      "grad_norm": 1.6850173549742617,
      "learning_rate": 0.00010478433887808696,
      "loss": 1.0225,
      "step": 6940
    },
    {
      "epoch": 0.5003243710805161,
      "grad_norm": 0.13943433170681396,
      "learning_rate": 0.00010476101833279681,
      "loss": 0.9356,
      "step": 6941
    },
    {
      "epoch": 0.500396453542853,
      "grad_norm": 0.1188989131563267,
      "learning_rate": 0.00010473769752798859,
      "loss": 1.0519,
      "step": 6942
    },
    {
      "epoch": 0.5004685360051899,
      "grad_norm": 0.21036688035328396,
      "learning_rate": 0.00010471437646493347,
      "loss": 1.0804,
      "step": 6943
    },
    {
      "epoch": 0.5005406184675268,
      "grad_norm": 0.22552893860852177,
      "learning_rate": 0.00010469105514490265,
      "loss": 1.0084,
      "step": 6944
    },
    {
      "epoch": 0.5006127009298638,
      "grad_norm": 0.11420045868781917,
      "learning_rate": 0.00010466773356916737,
      "loss": 1.0694,
      "step": 6945
    },
    {
      "epoch": 0.5006847833922007,
      "grad_norm": 0.11667411488218958,
      "learning_rate": 0.00010464441173899884,
      "loss": 0.9666,
      "step": 6946
    },
    {
      "epoch": 0.5007568658545376,
      "grad_norm": 0.12386583304851319,
      "learning_rate": 0.0001046210896556683,
      "loss": 1.0538,
      "step": 6947
    },
    {
      "epoch": 0.5008289483168745,
      "grad_norm": 0.3233609068474057,
      "learning_rate": 0.00010459776732044706,
      "loss": 1.021,
      "step": 6948
    },
    {
      "epoch": 0.5009010307792114,
      "grad_norm": 0.12917626504764979,
      "learning_rate": 0.00010457444473460635,
      "loss": 1.0415,
      "step": 6949
    },
    {
      "epoch": 0.5009731132415484,
      "grad_norm": 0.130328475226917,
      "learning_rate": 0.00010455112189941752,
      "loss": 1.0995,
      "step": 6950
    },
    {
      "epoch": 0.5010451957038853,
      "grad_norm": 0.14116376541116787,
      "learning_rate": 0.00010452779881615177,
      "loss": 1.0107,
      "step": 6951
    },
    {
      "epoch": 0.5011172781662222,
      "grad_norm": 0.20591237952084585,
      "learning_rate": 0.00010450447548608054,
      "loss": 1.0955,
      "step": 6952
    },
    {
      "epoch": 0.5011893606285591,
      "grad_norm": 0.12409790905031705,
      "learning_rate": 0.00010448115191047505,
      "loss": 1.0052,
      "step": 6953
    },
    {
      "epoch": 0.5012614430908959,
      "grad_norm": 0.1280059296440297,
      "learning_rate": 0.00010445782809060671,
      "loss": 1.0945,
      "step": 6954
    },
    {
      "epoch": 0.5013335255532329,
      "grad_norm": 0.12161387261530374,
      "learning_rate": 0.00010443450402774684,
      "loss": 1.0359,
      "step": 6955
    },
    {
      "epoch": 0.5014056080155698,
      "grad_norm": 0.235482665573265,
      "learning_rate": 0.00010441117972316688,
      "loss": 1.1188,
      "step": 6956
    },
    {
      "epoch": 0.5014776904779067,
      "grad_norm": 0.12959911062005966,
      "learning_rate": 0.00010438785517813809,
      "loss": 1.0067,
      "step": 6957
    },
    {
      "epoch": 0.5015497729402436,
      "grad_norm": 0.13682640112944489,
      "learning_rate": 0.00010436453039393199,
      "loss": 0.9787,
      "step": 6958
    },
    {
      "epoch": 0.5016218554025805,
      "grad_norm": 0.1616167994530399,
      "learning_rate": 0.0001043412053718199,
      "loss": 1.0583,
      "step": 6959
    },
    {
      "epoch": 0.5016939378649174,
      "grad_norm": 0.12077164178572078,
      "learning_rate": 0.00010431788011307333,
      "loss": 1.0215,
      "step": 6960
    },
    {
      "epoch": 0.5017660203272544,
      "grad_norm": 0.1299426404944069,
      "learning_rate": 0.00010429455461896363,
      "loss": 1.0284,
      "step": 6961
    },
    {
      "epoch": 0.5018381027895913,
      "grad_norm": 0.12252041218082281,
      "learning_rate": 0.00010427122889076228,
      "loss": 1.0699,
      "step": 6962
    },
    {
      "epoch": 0.5019101852519282,
      "grad_norm": 0.3638731925356462,
      "learning_rate": 0.00010424790292974076,
      "loss": 0.9999,
      "step": 6963
    },
    {
      "epoch": 0.5019822677142651,
      "grad_norm": 0.13327201849392464,
      "learning_rate": 0.00010422457673717052,
      "loss": 1.0265,
      "step": 6964
    },
    {
      "epoch": 0.502054350176602,
      "grad_norm": 0.14568107625568127,
      "learning_rate": 0.00010420125031432304,
      "loss": 1.1277,
      "step": 6965
    },
    {
      "epoch": 0.502126432638939,
      "grad_norm": 0.11736623803820519,
      "learning_rate": 0.00010417792366246986,
      "loss": 1.1191,
      "step": 6966
    },
    {
      "epoch": 0.5021985151012759,
      "grad_norm": 0.12911650515679468,
      "learning_rate": 0.00010415459678288244,
      "loss": 0.9483,
      "step": 6967
    },
    {
      "epoch": 0.5022705975636128,
      "grad_norm": 0.12999247978470177,
      "learning_rate": 0.00010413126967683233,
      "loss": 1.0821,
      "step": 6968
    },
    {
      "epoch": 0.5023426800259497,
      "grad_norm": 0.11952235264559755,
      "learning_rate": 0.00010410794234559107,
      "loss": 1.0808,
      "step": 6969
    },
    {
      "epoch": 0.5024147624882866,
      "grad_norm": 0.11850459223328148,
      "learning_rate": 0.00010408461479043021,
      "loss": 1.0393,
      "step": 6970
    },
    {
      "epoch": 0.5024868449506236,
      "grad_norm": 0.12659023036708977,
      "learning_rate": 0.00010406128701262128,
      "loss": 1.1426,
      "step": 6971
    },
    {
      "epoch": 0.5025589274129604,
      "grad_norm": 0.16322396953768703,
      "learning_rate": 0.00010403795901343592,
      "loss": 0.9728,
      "step": 6972
    },
    {
      "epoch": 0.5026310098752973,
      "grad_norm": 0.16079988920760113,
      "learning_rate": 0.00010401463079414562,
      "loss": 1.1301,
      "step": 6973
    },
    {
      "epoch": 0.5027030923376342,
      "grad_norm": 0.4471969021293219,
      "learning_rate": 0.00010399130235602206,
      "loss": 0.9917,
      "step": 6974
    },
    {
      "epoch": 0.5027751747999711,
      "grad_norm": 0.15449355918052943,
      "learning_rate": 0.00010396797370033678,
      "loss": 1.0108,
      "step": 6975
    },
    {
      "epoch": 0.5028472572623081,
      "grad_norm": 0.12459104301560553,
      "learning_rate": 0.00010394464482836146,
      "loss": 1.0287,
      "step": 6976
    },
    {
      "epoch": 0.502919339724645,
      "grad_norm": 0.12846324637573314,
      "learning_rate": 0.00010392131574136767,
      "loss": 1.0061,
      "step": 6977
    },
    {
      "epoch": 0.5029914221869819,
      "grad_norm": 0.19573225585681567,
      "learning_rate": 0.00010389798644062714,
      "loss": 1.0173,
      "step": 6978
    },
    {
      "epoch": 0.5030635046493188,
      "grad_norm": 0.117835910798688,
      "learning_rate": 0.00010387465692741142,
      "loss": 1.0459,
      "step": 6979
    },
    {
      "epoch": 0.5031355871116557,
      "grad_norm": 0.14183209435306934,
      "learning_rate": 0.00010385132720299231,
      "loss": 1.0773,
      "step": 6980
    },
    {
      "epoch": 0.5032076695739927,
      "grad_norm": 0.13610780385751062,
      "learning_rate": 0.00010382799726864135,
      "loss": 1.0187,
      "step": 6981
    },
    {
      "epoch": 0.5032797520363296,
      "grad_norm": 0.17370497389992012,
      "learning_rate": 0.00010380466712563033,
      "loss": 0.9973,
      "step": 6982
    },
    {
      "epoch": 0.5033518344986665,
      "grad_norm": 0.15730860287876433,
      "learning_rate": 0.0001037813367752309,
      "loss": 1.0473,
      "step": 6983
    },
    {
      "epoch": 0.5034239169610034,
      "grad_norm": 0.12798539664321776,
      "learning_rate": 0.00010375800621871478,
      "loss": 1.0295,
      "step": 6984
    },
    {
      "epoch": 0.5034959994233403,
      "grad_norm": 0.12756719867645097,
      "learning_rate": 0.00010373467545735371,
      "loss": 1.0548,
      "step": 6985
    },
    {
      "epoch": 0.5035680818856773,
      "grad_norm": 0.1618941019893019,
      "learning_rate": 0.00010371134449241941,
      "loss": 1.0179,
      "step": 6986
    },
    {
      "epoch": 0.5036401643480142,
      "grad_norm": 0.12673330072529654,
      "learning_rate": 0.00010368801332518365,
      "loss": 1.093,
      "step": 6987
    },
    {
      "epoch": 0.5037122468103511,
      "grad_norm": 0.13215284960501975,
      "learning_rate": 0.00010366468195691814,
      "loss": 1.0396,
      "step": 6988
    },
    {
      "epoch": 0.5037843292726879,
      "grad_norm": 0.11846704246283551,
      "learning_rate": 0.0001036413503888947,
      "loss": 1.0204,
      "step": 6989
    },
    {
      "epoch": 0.5038564117350248,
      "grad_norm": 0.26353184679658753,
      "learning_rate": 0.0001036180186223851,
      "loss": 1.034,
      "step": 6990
    },
    {
      "epoch": 0.5039284941973617,
      "grad_norm": 0.2815931518696399,
      "learning_rate": 0.00010359468665866108,
      "loss": 1.0361,
      "step": 6991
    },
    {
      "epoch": 0.5040005766596987,
      "grad_norm": 0.16554724413062882,
      "learning_rate": 0.00010357135449899449,
      "loss": 1.0207,
      "step": 6992
    },
    {
      "epoch": 0.5040726591220356,
      "grad_norm": 0.14453487986599034,
      "learning_rate": 0.00010354802214465715,
      "loss": 1.1192,
      "step": 6993
    },
    {
      "epoch": 0.5041447415843725,
      "grad_norm": 0.11683602177417143,
      "learning_rate": 0.00010352468959692083,
      "loss": 1.0068,
      "step": 6994
    },
    {
      "epoch": 0.5042168240467094,
      "grad_norm": 0.11372978336464956,
      "learning_rate": 0.00010350135685705742,
      "loss": 0.9949,
      "step": 6995
    },
    {
      "epoch": 0.5042889065090463,
      "grad_norm": 0.13880646900171478,
      "learning_rate": 0.00010347802392633874,
      "loss": 1.0837,
      "step": 6996
    },
    {
      "epoch": 0.5043609889713833,
      "grad_norm": 0.15719567224370148,
      "learning_rate": 0.00010345469080603663,
      "loss": 0.9697,
      "step": 6997
    },
    {
      "epoch": 0.5044330714337202,
      "grad_norm": 0.13804843256935875,
      "learning_rate": 0.00010343135749742296,
      "loss": 1.0524,
      "step": 6998
    },
    {
      "epoch": 0.5045051538960571,
      "grad_norm": 0.13864827443388256,
      "learning_rate": 0.00010340802400176962,
      "loss": 1.0224,
      "step": 6999
    },
    {
      "epoch": 0.504577236358394,
      "grad_norm": 0.16819669497783604,
      "learning_rate": 0.00010338469032034845,
      "loss": 1.0644,
      "step": 7000
    },
    {
      "epoch": 0.5046493188207309,
      "grad_norm": 0.12189152957765625,
      "learning_rate": 0.00010336135645443142,
      "loss": 1.1077,
      "step": 7001
    },
    {
      "epoch": 0.5047214012830679,
      "grad_norm": 0.15487524159514882,
      "learning_rate": 0.00010333802240529038,
      "loss": 1.0616,
      "step": 7002
    },
    {
      "epoch": 0.5047934837454048,
      "grad_norm": 0.12080075849483587,
      "learning_rate": 0.00010331468817419725,
      "loss": 0.9993,
      "step": 7003
    },
    {
      "epoch": 0.5048655662077417,
      "grad_norm": 0.2418867818473083,
      "learning_rate": 0.00010329135376242396,
      "loss": 1.1608,
      "step": 7004
    },
    {
      "epoch": 0.5049376486700786,
      "grad_norm": 0.1695204241383582,
      "learning_rate": 0.00010326801917124248,
      "loss": 1.0214,
      "step": 7005
    },
    {
      "epoch": 0.5050097311324154,
      "grad_norm": 0.1220421502956223,
      "learning_rate": 0.00010324468440192464,
      "loss": 1.0696,
      "step": 7006
    },
    {
      "epoch": 0.5050818135947523,
      "grad_norm": 1.724822035261205,
      "learning_rate": 0.00010322134945574253,
      "loss": 1.0727,
      "step": 7007
    },
    {
      "epoch": 0.5051538960570893,
      "grad_norm": 0.1227471114649947,
      "learning_rate": 0.00010319801433396802,
      "loss": 1.0706,
      "step": 7008
    },
    {
      "epoch": 0.5052259785194262,
      "grad_norm": 0.14125222791926964,
      "learning_rate": 0.00010317467903787317,
      "loss": 1.0181,
      "step": 7009
    },
    {
      "epoch": 0.5052980609817631,
      "grad_norm": 0.19541003275593638,
      "learning_rate": 0.00010315134356872986,
      "loss": 1.0279,
      "step": 7010
    },
    {
      "epoch": 0.5053701434441,
      "grad_norm": 0.12873202756494626,
      "learning_rate": 0.00010312800792781016,
      "loss": 1.0253,
      "step": 7011
    },
    {
      "epoch": 0.505442225906437,
      "grad_norm": 0.3923924972547679,
      "learning_rate": 0.00010310467211638603,
      "loss": 1.0448,
      "step": 7012
    },
    {
      "epoch": 0.5055143083687739,
      "grad_norm": 0.1390298322416189,
      "learning_rate": 0.00010308133613572952,
      "loss": 0.9822,
      "step": 7013
    },
    {
      "epoch": 0.5055863908311108,
      "grad_norm": 0.12007862322539255,
      "learning_rate": 0.0001030579999871126,
      "loss": 1.1058,
      "step": 7014
    },
    {
      "epoch": 0.5056584732934477,
      "grad_norm": 0.21075227744931938,
      "learning_rate": 0.00010303466367180735,
      "loss": 1.1491,
      "step": 7015
    },
    {
      "epoch": 0.5057305557557846,
      "grad_norm": 0.12235131081532337,
      "learning_rate": 0.00010301132719108576,
      "loss": 1.069,
      "step": 7016
    },
    {
      "epoch": 0.5058026382181215,
      "grad_norm": 0.11795585327699054,
      "learning_rate": 0.0001029879905462199,
      "loss": 1.1115,
      "step": 7017
    },
    {
      "epoch": 0.5058747206804585,
      "grad_norm": 0.1564755997057425,
      "learning_rate": 0.0001029646537384818,
      "loss": 1.0151,
      "step": 7018
    },
    {
      "epoch": 0.5059468031427954,
      "grad_norm": 0.8811267695236622,
      "learning_rate": 0.00010294131676914358,
      "loss": 1.0311,
      "step": 7019
    },
    {
      "epoch": 0.5060188856051323,
      "grad_norm": 0.11403199281309721,
      "learning_rate": 0.00010291797963947729,
      "loss": 1.0603,
      "step": 7020
    },
    {
      "epoch": 0.5060909680674692,
      "grad_norm": 0.5621741610050808,
      "learning_rate": 0.00010289464235075499,
      "loss": 1.0192,
      "step": 7021
    },
    {
      "epoch": 0.5061630505298061,
      "grad_norm": 0.11920358209809055,
      "learning_rate": 0.00010287130490424877,
      "loss": 0.9947,
      "step": 7022
    },
    {
      "epoch": 0.5062351329921431,
      "grad_norm": 0.12588886484796827,
      "learning_rate": 0.00010284796730123076,
      "loss": 1.0805,
      "step": 7023
    },
    {
      "epoch": 0.5063072154544799,
      "grad_norm": 0.14424676423989582,
      "learning_rate": 0.00010282462954297305,
      "loss": 1.0354,
      "step": 7024
    },
    {
      "epoch": 0.5063792979168168,
      "grad_norm": 0.13723647499571393,
      "learning_rate": 0.00010280129163074778,
      "loss": 1.066,
      "step": 7025
    },
    {
      "epoch": 0.5064513803791537,
      "grad_norm": 0.21784652483745193,
      "learning_rate": 0.00010277795356582702,
      "loss": 1.0181,
      "step": 7026
    },
    {
      "epoch": 0.5065234628414906,
      "grad_norm": 0.18756940835395644,
      "learning_rate": 0.00010275461534948298,
      "loss": 1.0574,
      "step": 7027
    },
    {
      "epoch": 0.5065955453038276,
      "grad_norm": 0.3831320144241045,
      "learning_rate": 0.00010273127698298772,
      "loss": 1.0336,
      "step": 7028
    },
    {
      "epoch": 0.5066676277661645,
      "grad_norm": 0.1181670398488053,
      "learning_rate": 0.00010270793846761347,
      "loss": 1.0944,
      "step": 7029
    },
    {
      "epoch": 0.5067397102285014,
      "grad_norm": 0.1235341017049881,
      "learning_rate": 0.0001026845998046323,
      "loss": 0.9864,
      "step": 7030
    },
    {
      "epoch": 0.5068117926908383,
      "grad_norm": 0.12336969386802071,
      "learning_rate": 0.00010266126099531648,
      "loss": 1.0585,
      "step": 7031
    },
    {
      "epoch": 0.5068838751531752,
      "grad_norm": 0.14618760997062233,
      "learning_rate": 0.00010263792204093807,
      "loss": 1.0345,
      "step": 7032
    },
    {
      "epoch": 0.5069559576155122,
      "grad_norm": 0.14228413413429317,
      "learning_rate": 0.00010261458294276935,
      "loss": 1.1156,
      "step": 7033
    },
    {
      "epoch": 0.5070280400778491,
      "grad_norm": 0.1824525409102259,
      "learning_rate": 0.00010259124370208243,
      "loss": 1.1221,
      "step": 7034
    },
    {
      "epoch": 0.507100122540186,
      "grad_norm": 0.12384233511606292,
      "learning_rate": 0.0001025679043201496,
      "loss": 1.0549,
      "step": 7035
    },
    {
      "epoch": 0.5071722050025229,
      "grad_norm": 0.11923066969129552,
      "learning_rate": 0.00010254456479824296,
      "loss": 1.078,
      "step": 7036
    },
    {
      "epoch": 0.5072442874648598,
      "grad_norm": 0.12548283000839883,
      "learning_rate": 0.0001025212251376348,
      "loss": 1.003,
      "step": 7037
    },
    {
      "epoch": 0.5073163699271968,
      "grad_norm": 0.23819675953757094,
      "learning_rate": 0.00010249788533959732,
      "loss": 0.9952,
      "step": 7038
    },
    {
      "epoch": 0.5073884523895337,
      "grad_norm": 0.13689564329375703,
      "learning_rate": 0.00010247454540540273,
      "loss": 1.0529,
      "step": 7039
    },
    {
      "epoch": 0.5074605348518706,
      "grad_norm": 0.11158705960978897,
      "learning_rate": 0.00010245120533632329,
      "loss": 1.0927,
      "step": 7040
    },
    {
      "epoch": 0.5075326173142074,
      "grad_norm": 0.12028293343829927,
      "learning_rate": 0.00010242786513363121,
      "loss": 0.9877,
      "step": 7041
    },
    {
      "epoch": 0.5076046997765443,
      "grad_norm": 0.1560261892401944,
      "learning_rate": 0.00010240452479859876,
      "loss": 1.0476,
      "step": 7042
    },
    {
      "epoch": 0.5076767822388812,
      "grad_norm": 0.13224756469617918,
      "learning_rate": 0.00010238118433249822,
      "loss": 1.0561,
      "step": 7043
    },
    {
      "epoch": 0.5077488647012182,
      "grad_norm": 0.15914215528605802,
      "learning_rate": 0.00010235784373660182,
      "loss": 1.0306,
      "step": 7044
    },
    {
      "epoch": 0.5078209471635551,
      "grad_norm": 0.12256777146715353,
      "learning_rate": 0.00010233450301218183,
      "loss": 1.0945,
      "step": 7045
    },
    {
      "epoch": 0.507893029625892,
      "grad_norm": 0.1636821372205985,
      "learning_rate": 0.00010231116216051055,
      "loss": 1.0321,
      "step": 7046
    },
    {
      "epoch": 0.5079651120882289,
      "grad_norm": 0.1257035961435484,
      "learning_rate": 0.00010228782118286028,
      "loss": 1.1028,
      "step": 7047
    },
    {
      "epoch": 0.5080371945505658,
      "grad_norm": 0.5570032616304792,
      "learning_rate": 0.00010226448008050325,
      "loss": 0.9881,
      "step": 7048
    },
    {
      "epoch": 0.5081092770129028,
      "grad_norm": 0.12881745342530312,
      "learning_rate": 0.00010224113885471183,
      "loss": 1.0954,
      "step": 7049
    },
    {
      "epoch": 0.5081813594752397,
      "grad_norm": 0.1438203055784519,
      "learning_rate": 0.00010221779750675824,
      "loss": 1.067,
      "step": 7050
    },
    {
      "epoch": 0.5082534419375766,
      "grad_norm": 0.15059198649702032,
      "learning_rate": 0.00010219445603791491,
      "loss": 1.0178,
      "step": 7051
    },
    {
      "epoch": 0.5083255243999135,
      "grad_norm": 0.11853282695435434,
      "learning_rate": 0.00010217111444945405,
      "loss": 0.9824,
      "step": 7052
    },
    {
      "epoch": 0.5083976068622504,
      "grad_norm": 0.10617651453767295,
      "learning_rate": 0.00010214777274264806,
      "loss": 1.053,
      "step": 7053
    },
    {
      "epoch": 0.5084696893245874,
      "grad_norm": 0.12768428381030805,
      "learning_rate": 0.0001021244309187692,
      "loss": 1.0613,
      "step": 7054
    },
    {
      "epoch": 0.5085417717869243,
      "grad_norm": 0.14012318458024622,
      "learning_rate": 0.0001021010889790899,
      "loss": 1.0715,
      "step": 7055
    },
    {
      "epoch": 0.5086138542492612,
      "grad_norm": 0.12205222079751528,
      "learning_rate": 0.00010207774692488238,
      "loss": 1.0231,
      "step": 7056
    },
    {
      "epoch": 0.5086859367115981,
      "grad_norm": 0.1129568519268077,
      "learning_rate": 0.00010205440475741915,
      "loss": 1.132,
      "step": 7057
    },
    {
      "epoch": 0.5087580191739349,
      "grad_norm": 0.10918711808748945,
      "learning_rate": 0.00010203106247797243,
      "loss": 1.013,
      "step": 7058
    },
    {
      "epoch": 0.5088301016362718,
      "grad_norm": 0.15642572656604312,
      "learning_rate": 0.00010200772008781465,
      "loss": 1.1249,
      "step": 7059
    },
    {
      "epoch": 0.5089021840986088,
      "grad_norm": 0.1301933893106079,
      "learning_rate": 0.00010198437758821817,
      "loss": 1.0266,
      "step": 7060
    },
    {
      "epoch": 0.5089742665609457,
      "grad_norm": 0.12141644164711027,
      "learning_rate": 0.00010196103498045535,
      "loss": 1.0026,
      "step": 7061
    },
    {
      "epoch": 0.5090463490232826,
      "grad_norm": 0.12311497385952286,
      "learning_rate": 0.00010193769226579857,
      "loss": 1.0436,
      "step": 7062
    },
    {
      "epoch": 0.5091184314856195,
      "grad_norm": 0.11849896221511656,
      "learning_rate": 0.00010191434944552022,
      "loss": 1.07,
      "step": 7063
    },
    {
      "epoch": 0.5091905139479564,
      "grad_norm": 0.15292449803933833,
      "learning_rate": 0.00010189100652089274,
      "loss": 1.075,
      "step": 7064
    },
    {
      "epoch": 0.5092625964102934,
      "grad_norm": 0.11054183556177721,
      "learning_rate": 0.00010186766349318846,
      "loss": 1.0645,
      "step": 7065
    },
    {
      "epoch": 0.5093346788726303,
      "grad_norm": 0.19493901150350898,
      "learning_rate": 0.00010184432036367981,
      "loss": 1.046,
      "step": 7066
    },
    {
      "epoch": 0.5094067613349672,
      "grad_norm": 0.11103129712701287,
      "learning_rate": 0.0001018209771336392,
      "loss": 1.0717,
      "step": 7067
    },
    {
      "epoch": 0.5094788437973041,
      "grad_norm": 0.6683348657973595,
      "learning_rate": 0.00010179763380433904,
      "loss": 1.1162,
      "step": 7068
    },
    {
      "epoch": 0.509550926259641,
      "grad_norm": 0.16043848618200388,
      "learning_rate": 0.00010177429037705177,
      "loss": 1.0986,
      "step": 7069
    },
    {
      "epoch": 0.509623008721978,
      "grad_norm": 0.10793218038741907,
      "learning_rate": 0.0001017509468530498,
      "loss": 1.0815,
      "step": 7070
    },
    {
      "epoch": 0.5096950911843149,
      "grad_norm": 0.10893805411306699,
      "learning_rate": 0.00010172760323360555,
      "loss": 1.0169,
      "step": 7071
    },
    {
      "epoch": 0.5097671736466518,
      "grad_norm": 0.12217048546852947,
      "learning_rate": 0.00010170425951999148,
      "loss": 1.0636,
      "step": 7072
    },
    {
      "epoch": 0.5098392561089887,
      "grad_norm": 0.12242041784805462,
      "learning_rate": 0.00010168091571348002,
      "loss": 1.1019,
      "step": 7073
    },
    {
      "epoch": 0.5099113385713256,
      "grad_norm": 0.1235571848105015,
      "learning_rate": 0.00010165757181534359,
      "loss": 1.072,
      "step": 7074
    },
    {
      "epoch": 0.5099834210336626,
      "grad_norm": 0.11401868903227405,
      "learning_rate": 0.0001016342278268547,
      "loss": 1.051,
      "step": 7075
    },
    {
      "epoch": 0.5100555034959994,
      "grad_norm": 0.1108917847890775,
      "learning_rate": 0.00010161088374928576,
      "loss": 1.0449,
      "step": 7076
    },
    {
      "epoch": 0.5101275859583363,
      "grad_norm": 0.5353291196478502,
      "learning_rate": 0.00010158753958390923,
      "loss": 0.9688,
      "step": 7077
    },
    {
      "epoch": 0.5101996684206732,
      "grad_norm": 0.1321724197482811,
      "learning_rate": 0.00010156419533199761,
      "loss": 1.0105,
      "step": 7078
    },
    {
      "epoch": 0.5102717508830101,
      "grad_norm": 0.20265753204174142,
      "learning_rate": 0.00010154085099482335,
      "loss": 1.0315,
      "step": 7079
    },
    {
      "epoch": 0.510343833345347,
      "grad_norm": 0.19657343220829415,
      "learning_rate": 0.00010151750657365895,
      "loss": 1.118,
      "step": 7080
    },
    {
      "epoch": 0.510415915807684,
      "grad_norm": 0.10992505541519744,
      "learning_rate": 0.0001014941620697768,
      "loss": 1.0582,
      "step": 7081
    },
    {
      "epoch": 0.5104879982700209,
      "grad_norm": 0.15958837475781834,
      "learning_rate": 0.00010147081748444951,
      "loss": 1.0818,
      "step": 7082
    },
    {
      "epoch": 0.5105600807323578,
      "grad_norm": 0.132748881328105,
      "learning_rate": 0.00010144747281894946,
      "loss": 1.0803,
      "step": 7083
    },
    {
      "epoch": 0.5106321631946947,
      "grad_norm": 0.11448110086404596,
      "learning_rate": 0.00010142412807454924,
      "loss": 1.0133,
      "step": 7084
    },
    {
      "epoch": 0.5107042456570317,
      "grad_norm": 0.11322800596769464,
      "learning_rate": 0.00010140078325252127,
      "loss": 1.0378,
      "step": 7085
    },
    {
      "epoch": 0.5107763281193686,
      "grad_norm": 0.10753194457861057,
      "learning_rate": 0.0001013774383541381,
      "loss": 0.9559,
      "step": 7086
    },
    {
      "epoch": 0.5108484105817055,
      "grad_norm": 0.35509820381591095,
      "learning_rate": 0.00010135409338067219,
      "loss": 1.1539,
      "step": 7087
    },
    {
      "epoch": 0.5109204930440424,
      "grad_norm": 0.11479646031656059,
      "learning_rate": 0.00010133074833339611,
      "loss": 1.1313,
      "step": 7088
    },
    {
      "epoch": 0.5109925755063793,
      "grad_norm": 0.17981485010613657,
      "learning_rate": 0.0001013074032135823,
      "loss": 1.0742,
      "step": 7089
    },
    {
      "epoch": 0.5110646579687163,
      "grad_norm": 0.1276608719869411,
      "learning_rate": 0.00010128405802250336,
      "loss": 1.1152,
      "step": 7090
    },
    {
      "epoch": 0.5111367404310532,
      "grad_norm": 0.18993678781541445,
      "learning_rate": 0.00010126071276143176,
      "loss": 1.0836,
      "step": 7091
    },
    {
      "epoch": 0.5112088228933901,
      "grad_norm": 0.1156552814076727,
      "learning_rate": 0.00010123736743164002,
      "loss": 0.9935,
      "step": 7092
    },
    {
      "epoch": 0.5112809053557269,
      "grad_norm": 0.145616576959741,
      "learning_rate": 0.0001012140220344007,
      "loss": 1.047,
      "step": 7093
    },
    {
      "epoch": 0.5113529878180638,
      "grad_norm": 0.13045824235038833,
      "learning_rate": 0.00010119067657098631,
      "loss": 1.0618,
      "step": 7094
    },
    {
      "epoch": 0.5114250702804007,
      "grad_norm": 0.1176815687866234,
      "learning_rate": 0.0001011673310426694,
      "loss": 1.0516,
      "step": 7095
    },
    {
      "epoch": 0.5114971527427377,
      "grad_norm": 0.13989055996575758,
      "learning_rate": 0.0001011439854507225,
      "loss": 1.0551,
      "step": 7096
    },
    {
      "epoch": 0.5115692352050746,
      "grad_norm": 0.12494338204592122,
      "learning_rate": 0.00010112063979641815,
      "loss": 1.0196,
      "step": 7097
    },
    {
      "epoch": 0.5116413176674115,
      "grad_norm": 0.1121703349473662,
      "learning_rate": 0.00010109729408102891,
      "loss": 1.1166,
      "step": 7098
    },
    {
      "epoch": 0.5117134001297484,
      "grad_norm": 0.13062913607357687,
      "learning_rate": 0.00010107394830582732,
      "loss": 1.0059,
      "step": 7099
    },
    {
      "epoch": 0.5117854825920853,
      "grad_norm": 2.517729736931236,
      "learning_rate": 0.00010105060247208595,
      "loss": 1.1217,
      "step": 7100
    },
    {
      "epoch": 0.5118575650544223,
      "grad_norm": 0.2068619614730265,
      "learning_rate": 0.00010102725658107737,
      "loss": 1.1006,
      "step": 7101
    },
    {
      "epoch": 0.5119296475167592,
      "grad_norm": 0.18169618110298819,
      "learning_rate": 0.00010100391063407411,
      "loss": 0.9898,
      "step": 7102
    },
    {
      "epoch": 0.5120017299790961,
      "grad_norm": 0.12709572920885473,
      "learning_rate": 0.00010098056463234869,
      "loss": 0.9875,
      "step": 7103
    },
    {
      "epoch": 0.512073812441433,
      "grad_norm": 0.12434331272517894,
      "learning_rate": 0.00010095721857717378,
      "loss": 1.0333,
      "step": 7104
    },
    {
      "epoch": 0.5121458949037699,
      "grad_norm": 0.168047355315689,
      "learning_rate": 0.00010093387246982184,
      "loss": 1.0174,
      "step": 7105
    },
    {
      "epoch": 0.5122179773661069,
      "grad_norm": 0.14055584579351132,
      "learning_rate": 0.00010091052631156555,
      "loss": 1.1089,
      "step": 7106
    },
    {
      "epoch": 0.5122900598284438,
      "grad_norm": 0.31607379892799103,
      "learning_rate": 0.00010088718010367739,
      "loss": 1.0147,
      "step": 7107
    },
    {
      "epoch": 0.5123621422907807,
      "grad_norm": 0.2490676487811558,
      "learning_rate": 0.00010086383384743001,
      "loss": 1.0955,
      "step": 7108
    },
    {
      "epoch": 0.5124342247531176,
      "grad_norm": 0.11663754580132192,
      "learning_rate": 0.00010084048754409592,
      "loss": 1.0928,
      "step": 7109
    },
    {
      "epoch": 0.5125063072154544,
      "grad_norm": 1.0438027862094381,
      "learning_rate": 0.00010081714119494779,
      "loss": 1.1148,
      "step": 7110
    },
    {
      "epoch": 0.5125783896777913,
      "grad_norm": 0.11599662069585298,
      "learning_rate": 0.0001007937948012581,
      "loss": 1.0674,
      "step": 7111
    },
    {
      "epoch": 0.5126504721401283,
      "grad_norm": 0.12570342098934179,
      "learning_rate": 0.00010077044836429954,
      "loss": 1.0497,
      "step": 7112
    },
    {
      "epoch": 0.5127225546024652,
      "grad_norm": 0.3706745820185091,
      "learning_rate": 0.00010074710188534462,
      "loss": 1.0535,
      "step": 7113
    },
    {
      "epoch": 0.5127946370648021,
      "grad_norm": 0.11480165941688138,
      "learning_rate": 0.00010072375536566599,
      "loss": 1.1007,
      "step": 7114
    },
    {
      "epoch": 0.512866719527139,
      "grad_norm": 0.20914612307385663,
      "learning_rate": 0.0001007004088065362,
      "loss": 0.9839,
      "step": 7115
    },
    {
      "epoch": 0.512938801989476,
      "grad_norm": 0.6816383909831469,
      "learning_rate": 0.00010067706220922785,
      "loss": 1.0532,
      "step": 7116
    },
    {
      "epoch": 0.5130108844518129,
      "grad_norm": 0.1325885166161065,
      "learning_rate": 0.00010065371557501358,
      "loss": 1.011,
      "step": 7117
    },
    {
      "epoch": 0.5130829669141498,
      "grad_norm": 0.10967319515971778,
      "learning_rate": 0.00010063036890516593,
      "loss": 1.056,
      "step": 7118
    },
    {
      "epoch": 0.5131550493764867,
      "grad_norm": 0.11924326030882249,
      "learning_rate": 0.00010060702220095756,
      "loss": 1.0034,
      "step": 7119
    },
    {
      "epoch": 0.5132271318388236,
      "grad_norm": 0.4696966568127836,
      "learning_rate": 0.00010058367546366105,
      "loss": 1.0579,
      "step": 7120
    },
    {
      "epoch": 0.5132992143011605,
      "grad_norm": 0.1491488021227365,
      "learning_rate": 0.000100560328694549,
      "loss": 1.0078,
      "step": 7121
    },
    {
      "epoch": 0.5133712967634975,
      "grad_norm": 1.0179317764371707,
      "learning_rate": 0.00010053698189489402,
      "loss": 1.035,
      "step": 7122
    },
    {
      "epoch": 0.5134433792258344,
      "grad_norm": 0.2583031739942567,
      "learning_rate": 0.00010051363506596872,
      "loss": 1.079,
      "step": 7123
    },
    {
      "epoch": 0.5135154616881713,
      "grad_norm": 0.13253024175625638,
      "learning_rate": 0.00010049028820904574,
      "loss": 1.0953,
      "step": 7124
    },
    {
      "epoch": 0.5135875441505082,
      "grad_norm": 1.0797952280060603,
      "learning_rate": 0.00010046694132539762,
      "loss": 0.946,
      "step": 7125
    },
    {
      "epoch": 0.5136596266128451,
      "grad_norm": 0.127884918690722,
      "learning_rate": 0.00010044359441629707,
      "loss": 1.0624,
      "step": 7126
    },
    {
      "epoch": 0.5137317090751821,
      "grad_norm": 0.17890469055615058,
      "learning_rate": 0.00010042024748301661,
      "loss": 1.1195,
      "step": 7127
    },
    {
      "epoch": 0.5138037915375189,
      "grad_norm": 0.12210012202202543,
      "learning_rate": 0.00010039690052682895,
      "loss": 1.0888,
      "step": 7128
    },
    {
      "epoch": 0.5138758739998558,
      "grad_norm": 0.12106189508036684,
      "learning_rate": 0.00010037355354900661,
      "loss": 1.01,
      "step": 7129
    },
    {
      "epoch": 0.5139479564621927,
      "grad_norm": 0.1316299913678933,
      "learning_rate": 0.0001003502065508223,
      "loss": 1.0567,
      "step": 7130
    },
    {
      "epoch": 0.5140200389245296,
      "grad_norm": 0.23705076261020142,
      "learning_rate": 0.00010032685953354857,
      "loss": 1.0688,
      "step": 7131
    },
    {
      "epoch": 0.5140921213868666,
      "grad_norm": 0.13549151692662098,
      "learning_rate": 0.0001003035124984581,
      "loss": 1.0738,
      "step": 7132
    },
    {
      "epoch": 0.5141642038492035,
      "grad_norm": 0.15283106088480572,
      "learning_rate": 0.00010028016544682346,
      "loss": 1.0418,
      "step": 7133
    },
    {
      "epoch": 0.5142362863115404,
      "grad_norm": 0.21930639625697282,
      "learning_rate": 0.00010025681837991731,
      "loss": 1.0647,
      "step": 7134
    },
    {
      "epoch": 0.5143083687738773,
      "grad_norm": 0.11931232328328696,
      "learning_rate": 0.00010023347129901226,
      "loss": 1.0706,
      "step": 7135
    },
    {
      "epoch": 0.5143804512362142,
      "grad_norm": 0.13516640807660935,
      "learning_rate": 0.00010021012420538094,
      "loss": 1.1197,
      "step": 7136
    },
    {
      "epoch": 0.5144525336985512,
      "grad_norm": 0.1549233013986965,
      "learning_rate": 0.00010018677710029596,
      "loss": 1.114,
      "step": 7137
    },
    {
      "epoch": 0.5145246161608881,
      "grad_norm": 0.5339811871602617,
      "learning_rate": 0.00010016342998502997,
      "loss": 1.0183,
      "step": 7138
    },
    {
      "epoch": 0.514596698623225,
      "grad_norm": 0.12876447200030142,
      "learning_rate": 0.00010014008286085556,
      "loss": 1.1045,
      "step": 7139
    },
    {
      "epoch": 0.5146687810855619,
      "grad_norm": 0.12127477199712122,
      "learning_rate": 0.00010011673572904541,
      "loss": 1.0263,
      "step": 7140
    },
    {
      "epoch": 0.5147408635478988,
      "grad_norm": 0.11219821621450599,
      "learning_rate": 0.00010009338859087209,
      "loss": 1.0776,
      "step": 7141
    },
    {
      "epoch": 0.5148129460102358,
      "grad_norm": 0.13751696770530178,
      "learning_rate": 0.00010007004144760828,
      "loss": 0.9735,
      "step": 7142
    },
    {
      "epoch": 0.5148850284725727,
      "grad_norm": 0.12164109630935675,
      "learning_rate": 0.00010004669430052656,
      "loss": 1.0424,
      "step": 7143
    },
    {
      "epoch": 0.5149571109349096,
      "grad_norm": 0.37242638342843476,
      "learning_rate": 0.00010002334715089961,
      "loss": 1.0575,
      "step": 7144
    },
    {
      "epoch": 0.5150291933972464,
      "grad_norm": 0.11129442498759959,
      "learning_rate": 0.0001,
      "loss": 1.0251,
      "step": 7145
    },
    {
      "epoch": 0.5151012758595833,
      "grad_norm": 0.13730568204820934,
      "learning_rate": 9.997665284910042e-05,
      "loss": 1.0563,
      "step": 7146
    },
    {
      "epoch": 0.5151733583219202,
      "grad_norm": 0.13501843171798453,
      "learning_rate": 9.995330569947348e-05,
      "loss": 1.0367,
      "step": 7147
    },
    {
      "epoch": 0.5152454407842572,
      "grad_norm": 0.12965383885086312,
      "learning_rate": 9.992995855239174e-05,
      "loss": 1.0263,
      "step": 7148
    },
    {
      "epoch": 0.5153175232465941,
      "grad_norm": 0.9687056214250568,
      "learning_rate": 9.990661140912793e-05,
      "loss": 1.0036,
      "step": 7149
    },
    {
      "epoch": 0.515389605708931,
      "grad_norm": 0.12368776055587465,
      "learning_rate": 9.988326427095463e-05,
      "loss": 1.0833,
      "step": 7150
    },
    {
      "epoch": 0.5154616881712679,
      "grad_norm": 0.40643433383853356,
      "learning_rate": 9.985991713914448e-05,
      "loss": 1.0291,
      "step": 7151
    },
    {
      "epoch": 0.5155337706336048,
      "grad_norm": 0.19433762712730468,
      "learning_rate": 9.983657001497005e-05,
      "loss": 1.0461,
      "step": 7152
    },
    {
      "epoch": 0.5156058530959418,
      "grad_norm": 0.1287841318572899,
      "learning_rate": 9.981322289970407e-05,
      "loss": 1.0054,
      "step": 7153
    },
    {
      "epoch": 0.5156779355582787,
      "grad_norm": 0.13235728989584145,
      "learning_rate": 9.978987579461908e-05,
      "loss": 1.0718,
      "step": 7154
    },
    {
      "epoch": 0.5157500180206156,
      "grad_norm": 0.11645979945293682,
      "learning_rate": 9.976652870098777e-05,
      "loss": 1.0543,
      "step": 7155
    },
    {
      "epoch": 0.5158221004829525,
      "grad_norm": 0.4651714579950913,
      "learning_rate": 9.974318162008273e-05,
      "loss": 1.0826,
      "step": 7156
    },
    {
      "epoch": 0.5158941829452894,
      "grad_norm": 0.15591239258998088,
      "learning_rate": 9.971983455317655e-05,
      "loss": 1.1371,
      "step": 7157
    },
    {
      "epoch": 0.5159662654076264,
      "grad_norm": 0.1517175518051269,
      "learning_rate": 9.969648750154192e-05,
      "loss": 1.0702,
      "step": 7158
    },
    {
      "epoch": 0.5160383478699633,
      "grad_norm": 0.11706480535198133,
      "learning_rate": 9.967314046645145e-05,
      "loss": 1.0705,
      "step": 7159
    },
    {
      "epoch": 0.5161104303323002,
      "grad_norm": 0.18527924653436187,
      "learning_rate": 9.964979344917775e-05,
      "loss": 1.0551,
      "step": 7160
    },
    {
      "epoch": 0.5161825127946371,
      "grad_norm": 0.1879161550066925,
      "learning_rate": 9.96264464509934e-05,
      "loss": 1.0378,
      "step": 7161
    },
    {
      "epoch": 0.5162545952569739,
      "grad_norm": 0.1291898186122468,
      "learning_rate": 9.960309947317106e-05,
      "loss": 1.0669,
      "step": 7162
    },
    {
      "epoch": 0.5163266777193108,
      "grad_norm": 0.36594945600362955,
      "learning_rate": 9.957975251698341e-05,
      "loss": 1.0476,
      "step": 7163
    },
    {
      "epoch": 0.5163987601816478,
      "grad_norm": 0.16113211767343374,
      "learning_rate": 9.955640558370297e-05,
      "loss": 0.9631,
      "step": 7164
    },
    {
      "epoch": 0.5164708426439847,
      "grad_norm": 0.12800838470502765,
      "learning_rate": 9.953305867460238e-05,
      "loss": 0.9358,
      "step": 7165
    },
    {
      "epoch": 0.5165429251063216,
      "grad_norm": 0.16780907830528433,
      "learning_rate": 9.950971179095429e-05,
      "loss": 1.0107,
      "step": 7166
    },
    {
      "epoch": 0.5166150075686585,
      "grad_norm": 0.12376379287810392,
      "learning_rate": 9.94863649340313e-05,
      "loss": 1.0335,
      "step": 7167
    },
    {
      "epoch": 0.5166870900309954,
      "grad_norm": 0.11195413862514196,
      "learning_rate": 9.9463018105106e-05,
      "loss": 1.0174,
      "step": 7168
    },
    {
      "epoch": 0.5167591724933324,
      "grad_norm": 0.11940673942706506,
      "learning_rate": 9.943967130545105e-05,
      "loss": 1.1121,
      "step": 7169
    },
    {
      "epoch": 0.5168312549556693,
      "grad_norm": 0.19695837548297646,
      "learning_rate": 9.941632453633896e-05,
      "loss": 1.1866,
      "step": 7170
    },
    {
      "epoch": 0.5169033374180062,
      "grad_norm": 0.12706248837566128,
      "learning_rate": 9.939297779904246e-05,
      "loss": 1.0581,
      "step": 7171
    },
    {
      "epoch": 0.5169754198803431,
      "grad_norm": 0.12475241135532317,
      "learning_rate": 9.936963109483408e-05,
      "loss": 1.0575,
      "step": 7172
    },
    {
      "epoch": 0.51704750234268,
      "grad_norm": 0.12430369807002696,
      "learning_rate": 9.934628442498649e-05,
      "loss": 1.0108,
      "step": 7173
    },
    {
      "epoch": 0.517119584805017,
      "grad_norm": 0.1371362175966942,
      "learning_rate": 9.932293779077216e-05,
      "loss": 1.1382,
      "step": 7174
    },
    {
      "epoch": 0.5171916672673539,
      "grad_norm": 0.3483253072996132,
      "learning_rate": 9.929959119346384e-05,
      "loss": 1.0161,
      "step": 7175
    },
    {
      "epoch": 0.5172637497296908,
      "grad_norm": 0.11623018954872347,
      "learning_rate": 9.927624463433405e-05,
      "loss": 1.0357,
      "step": 7176
    },
    {
      "epoch": 0.5173358321920277,
      "grad_norm": 0.1414157911534523,
      "learning_rate": 9.925289811465542e-05,
      "loss": 1.0603,
      "step": 7177
    },
    {
      "epoch": 0.5174079146543646,
      "grad_norm": 0.11866623156814561,
      "learning_rate": 9.92295516357005e-05,
      "loss": 1.0199,
      "step": 7178
    },
    {
      "epoch": 0.5174799971167015,
      "grad_norm": 0.14318415028109838,
      "learning_rate": 9.92062051987419e-05,
      "loss": 1.0408,
      "step": 7179
    },
    {
      "epoch": 0.5175520795790384,
      "grad_norm": 0.11995529010986672,
      "learning_rate": 9.918285880505223e-05,
      "loss": 1.0384,
      "step": 7180
    },
    {
      "epoch": 0.5176241620413753,
      "grad_norm": 0.1307613157051955,
      "learning_rate": 9.91595124559041e-05,
      "loss": 1.1234,
      "step": 7181
    },
    {
      "epoch": 0.5176962445037122,
      "grad_norm": 0.11283996411770696,
      "learning_rate": 9.913616615257004e-05,
      "loss": 1.1028,
      "step": 7182
    },
    {
      "epoch": 0.5177683269660491,
      "grad_norm": 0.5471312105512993,
      "learning_rate": 9.911281989632261e-05,
      "loss": 0.9875,
      "step": 7183
    },
    {
      "epoch": 0.517840409428386,
      "grad_norm": 0.12309577787492647,
      "learning_rate": 9.908947368843446e-05,
      "loss": 1.015,
      "step": 7184
    },
    {
      "epoch": 0.517912491890723,
      "grad_norm": 0.13844406231012574,
      "learning_rate": 9.906612753017818e-05,
      "loss": 1.0379,
      "step": 7185
    },
    {
      "epoch": 0.5179845743530599,
      "grad_norm": 0.11892345454447685,
      "learning_rate": 9.904278142282627e-05,
      "loss": 1.0098,
      "step": 7186
    },
    {
      "epoch": 0.5180566568153968,
      "grad_norm": 0.12809729217796237,
      "learning_rate": 9.901943536765131e-05,
      "loss": 1.1286,
      "step": 7187
    },
    {
      "epoch": 0.5181287392777337,
      "grad_norm": 0.12553882843318806,
      "learning_rate": 9.899608936592591e-05,
      "loss": 0.9779,
      "step": 7188
    },
    {
      "epoch": 0.5182008217400706,
      "grad_norm": 0.10623501983554556,
      "learning_rate": 9.897274341892267e-05,
      "loss": 1.0593,
      "step": 7189
    },
    {
      "epoch": 0.5182729042024076,
      "grad_norm": 0.4687789231441341,
      "learning_rate": 9.894939752791406e-05,
      "loss": 1.1067,
      "step": 7190
    },
    {
      "epoch": 0.5183449866647445,
      "grad_norm": 0.13684995253685917,
      "learning_rate": 9.892605169417271e-05,
      "loss": 1.1001,
      "step": 7191
    },
    {
      "epoch": 0.5184170691270814,
      "grad_norm": 0.12552763961364752,
      "learning_rate": 9.890270591897109e-05,
      "loss": 1.0017,
      "step": 7192
    },
    {
      "epoch": 0.5184891515894183,
      "grad_norm": 0.1492842998092864,
      "learning_rate": 9.887936020358187e-05,
      "loss": 1.0851,
      "step": 7193
    },
    {
      "epoch": 0.5185612340517552,
      "grad_norm": 0.1679527405161091,
      "learning_rate": 9.885601454927753e-05,
      "loss": 0.9908,
      "step": 7194
    },
    {
      "epoch": 0.5186333165140922,
      "grad_norm": 0.22462053639381654,
      "learning_rate": 9.883266895733066e-05,
      "loss": 1.1048,
      "step": 7195
    },
    {
      "epoch": 0.5187053989764291,
      "grad_norm": 0.11962627744445203,
      "learning_rate": 9.880932342901368e-05,
      "loss": 1.0926,
      "step": 7196
    },
    {
      "epoch": 0.5187774814387659,
      "grad_norm": 0.127117262234405,
      "learning_rate": 9.878597796559933e-05,
      "loss": 1.0587,
      "step": 7197
    },
    {
      "epoch": 0.5188495639011028,
      "grad_norm": 0.11541005247576366,
      "learning_rate": 9.876263256836e-05,
      "loss": 1.135,
      "step": 7198
    },
    {
      "epoch": 0.5189216463634397,
      "grad_norm": 0.11982092485353468,
      "learning_rate": 9.873928723856829e-05,
      "loss": 1.0527,
      "step": 7199
    },
    {
      "epoch": 0.5189937288257767,
      "grad_norm": 0.36883587026535913,
      "learning_rate": 9.871594197749668e-05,
      "loss": 1.1699,
      "step": 7200
    },
    {
      "epoch": 0.5190658112881136,
      "grad_norm": 0.11848673642912687,
      "learning_rate": 9.86925967864177e-05,
      "loss": 0.9799,
      "step": 7201
    },
    {
      "epoch": 0.5191378937504505,
      "grad_norm": 0.5560447064966074,
      "learning_rate": 9.866925166660391e-05,
      "loss": 1.0922,
      "step": 7202
    },
    {
      "epoch": 0.5192099762127874,
      "grad_norm": 0.21118470355952546,
      "learning_rate": 9.864590661932783e-05,
      "loss": 1.002,
      "step": 7203
    },
    {
      "epoch": 0.5192820586751243,
      "grad_norm": 1.1108563422209863,
      "learning_rate": 9.862256164586193e-05,
      "loss": 1.0071,
      "step": 7204
    },
    {
      "epoch": 0.5193541411374613,
      "grad_norm": 0.7715352834896999,
      "learning_rate": 9.859921674747873e-05,
      "loss": 1.0662,
      "step": 7205
    },
    {
      "epoch": 0.5194262235997982,
      "grad_norm": 0.15466106214453199,
      "learning_rate": 9.857587192545077e-05,
      "loss": 1.0615,
      "step": 7206
    },
    {
      "epoch": 0.5194983060621351,
      "grad_norm": 0.2606770728546372,
      "learning_rate": 9.855252718105055e-05,
      "loss": 1.0006,
      "step": 7207
    },
    {
      "epoch": 0.519570388524472,
      "grad_norm": 0.1521042437286639,
      "learning_rate": 9.852918251555052e-05,
      "loss": 1.0105,
      "step": 7208
    },
    {
      "epoch": 0.5196424709868089,
      "grad_norm": 0.12876561467026737,
      "learning_rate": 9.850583793022319e-05,
      "loss": 1.0355,
      "step": 7209
    },
    {
      "epoch": 0.5197145534491459,
      "grad_norm": 0.14642378674085374,
      "learning_rate": 9.848249342634108e-05,
      "loss": 1.0159,
      "step": 7210
    },
    {
      "epoch": 0.5197866359114828,
      "grad_norm": 0.17320865835105834,
      "learning_rate": 9.845914900517667e-05,
      "loss": 1.115,
      "step": 7211
    },
    {
      "epoch": 0.5198587183738197,
      "grad_norm": 0.21698747957614245,
      "learning_rate": 9.843580466800241e-05,
      "loss": 1.0718,
      "step": 7212
    },
    {
      "epoch": 0.5199308008361566,
      "grad_norm": 0.13337433442846933,
      "learning_rate": 9.841246041609078e-05,
      "loss": 1.0615,
      "step": 7213
    },
    {
      "epoch": 0.5200028832984934,
      "grad_norm": 0.11573228776593165,
      "learning_rate": 9.838911625071424e-05,
      "loss": 1.0713,
      "step": 7214
    },
    {
      "epoch": 0.5200749657608303,
      "grad_norm": 0.12291381803766416,
      "learning_rate": 9.836577217314533e-05,
      "loss": 1.0637,
      "step": 7215
    },
    {
      "epoch": 0.5201470482231673,
      "grad_norm": 0.27752024927389995,
      "learning_rate": 9.834242818465642e-05,
      "loss": 1.085,
      "step": 7216
    },
    {
      "epoch": 0.5202191306855042,
      "grad_norm": 0.1988775792742282,
      "learning_rate": 9.831908428652002e-05,
      "loss": 1.0804,
      "step": 7217
    },
    {
      "epoch": 0.5202912131478411,
      "grad_norm": 0.13113672041158775,
      "learning_rate": 9.829574048000852e-05,
      "loss": 1.0505,
      "step": 7218
    },
    {
      "epoch": 0.520363295610178,
      "grad_norm": 0.308995528543256,
      "learning_rate": 9.827239676639448e-05,
      "loss": 1.1367,
      "step": 7219
    },
    {
      "epoch": 0.5204353780725149,
      "grad_norm": 0.12546182822341734,
      "learning_rate": 9.824905314695023e-05,
      "loss": 1.0679,
      "step": 7220
    },
    {
      "epoch": 0.5205074605348519,
      "grad_norm": 0.13135410958424748,
      "learning_rate": 9.822570962294826e-05,
      "loss": 0.9914,
      "step": 7221
    },
    {
      "epoch": 0.5205795429971888,
      "grad_norm": 0.10937504861658941,
      "learning_rate": 9.8202366195661e-05,
      "loss": 1.0485,
      "step": 7222
    },
    {
      "epoch": 0.5206516254595257,
      "grad_norm": 0.11604778066440624,
      "learning_rate": 9.817902286636082e-05,
      "loss": 1.1032,
      "step": 7223
    },
    {
      "epoch": 0.5207237079218626,
      "grad_norm": 0.11737156103467397,
      "learning_rate": 9.815567963632022e-05,
      "loss": 1.0079,
      "step": 7224
    },
    {
      "epoch": 0.5207957903841995,
      "grad_norm": 0.12071029367626118,
      "learning_rate": 9.813233650681157e-05,
      "loss": 1.0633,
      "step": 7225
    },
    {
      "epoch": 0.5208678728465365,
      "grad_norm": 0.4652209336275358,
      "learning_rate": 9.810899347910732e-05,
      "loss": 1.0214,
      "step": 7226
    },
    {
      "epoch": 0.5209399553088734,
      "grad_norm": 0.13673263315982476,
      "learning_rate": 9.808565055447979e-05,
      "loss": 1.0617,
      "step": 7227
    },
    {
      "epoch": 0.5210120377712103,
      "grad_norm": 0.12270857340260946,
      "learning_rate": 9.806230773420145e-05,
      "loss": 1.0387,
      "step": 7228
    },
    {
      "epoch": 0.5210841202335472,
      "grad_norm": 0.11416674071383554,
      "learning_rate": 9.803896501954469e-05,
      "loss": 0.9975,
      "step": 7229
    },
    {
      "epoch": 0.5211562026958841,
      "grad_norm": 0.10327022200265272,
      "learning_rate": 9.801562241178188e-05,
      "loss": 0.9623,
      "step": 7230
    },
    {
      "epoch": 0.521228285158221,
      "grad_norm": 0.252646777363442,
      "learning_rate": 9.799227991218537e-05,
      "loss": 1.1218,
      "step": 7231
    },
    {
      "epoch": 0.5213003676205579,
      "grad_norm": 0.12183400263673037,
      "learning_rate": 9.796893752202758e-05,
      "loss": 1.063,
      "step": 7232
    },
    {
      "epoch": 0.5213724500828948,
      "grad_norm": 0.13571028897702042,
      "learning_rate": 9.794559524258088e-05,
      "loss": 1.0816,
      "step": 7233
    },
    {
      "epoch": 0.5214445325452317,
      "grad_norm": 0.1115964096402035,
      "learning_rate": 9.792225307511763e-05,
      "loss": 1.0168,
      "step": 7234
    },
    {
      "epoch": 0.5215166150075686,
      "grad_norm": 0.12319560660657174,
      "learning_rate": 9.789891102091015e-05,
      "loss": 1.0288,
      "step": 7235
    },
    {
      "epoch": 0.5215886974699055,
      "grad_norm": 0.11115831640525561,
      "learning_rate": 9.78755690812308e-05,
      "loss": 1.0133,
      "step": 7236
    },
    {
      "epoch": 0.5216607799322425,
      "grad_norm": 0.12810921030722083,
      "learning_rate": 9.785222725735197e-05,
      "loss": 0.9804,
      "step": 7237
    },
    {
      "epoch": 0.5217328623945794,
      "grad_norm": 0.11428571649984345,
      "learning_rate": 9.782888555054598e-05,
      "loss": 1.0919,
      "step": 7238
    },
    {
      "epoch": 0.5218049448569163,
      "grad_norm": 0.11433395867264487,
      "learning_rate": 9.780554396208514e-05,
      "loss": 1.0832,
      "step": 7239
    },
    {
      "epoch": 0.5218770273192532,
      "grad_norm": 0.11515752191047014,
      "learning_rate": 9.778220249324174e-05,
      "loss": 1.0821,
      "step": 7240
    },
    {
      "epoch": 0.5219491097815901,
      "grad_norm": 0.11933059598590534,
      "learning_rate": 9.77588611452882e-05,
      "loss": 1.0467,
      "step": 7241
    },
    {
      "epoch": 0.5220211922439271,
      "grad_norm": 0.10622511906185571,
      "learning_rate": 9.773551991949678e-05,
      "loss": 1.0073,
      "step": 7242
    },
    {
      "epoch": 0.522093274706264,
      "grad_norm": 0.12220570779422198,
      "learning_rate": 9.771217881713977e-05,
      "loss": 1.0473,
      "step": 7243
    },
    {
      "epoch": 0.5221653571686009,
      "grad_norm": 0.19321728425151152,
      "learning_rate": 9.768883783948948e-05,
      "loss": 1.0844,
      "step": 7244
    },
    {
      "epoch": 0.5222374396309378,
      "grad_norm": 0.13054879848461412,
      "learning_rate": 9.766549698781818e-05,
      "loss": 1.1048,
      "step": 7245
    },
    {
      "epoch": 0.5223095220932747,
      "grad_norm": 0.15498181134445008,
      "learning_rate": 9.764215626339822e-05,
      "loss": 1.0091,
      "step": 7246
    },
    {
      "epoch": 0.5223816045556117,
      "grad_norm": 0.12347915997693483,
      "learning_rate": 9.76188156675018e-05,
      "loss": 1.0413,
      "step": 7247
    },
    {
      "epoch": 0.5224536870179486,
      "grad_norm": 0.11771055810218746,
      "learning_rate": 9.759547520140128e-05,
      "loss": 1.0675,
      "step": 7248
    },
    {
      "epoch": 0.5225257694802854,
      "grad_norm": 0.09488807994191917,
      "learning_rate": 9.757213486636879e-05,
      "loss": 0.9807,
      "step": 7249
    },
    {
      "epoch": 0.5225978519426223,
      "grad_norm": 0.12862263162131488,
      "learning_rate": 9.754879466367675e-05,
      "loss": 1.1135,
      "step": 7250
    },
    {
      "epoch": 0.5226699344049592,
      "grad_norm": 0.15079047598081538,
      "learning_rate": 9.752545459459729e-05,
      "loss": 1.0045,
      "step": 7251
    },
    {
      "epoch": 0.5227420168672962,
      "grad_norm": 0.16839353140087204,
      "learning_rate": 9.750211466040272e-05,
      "loss": 1.0206,
      "step": 7252
    },
    {
      "epoch": 0.5228140993296331,
      "grad_norm": 0.12464948283223677,
      "learning_rate": 9.747877486236518e-05,
      "loss": 1.037,
      "step": 7253
    },
    {
      "epoch": 0.52288618179197,
      "grad_norm": 0.10450939136336894,
      "learning_rate": 9.745543520175705e-05,
      "loss": 0.9843,
      "step": 7254
    },
    {
      "epoch": 0.5229582642543069,
      "grad_norm": 0.19589057919967814,
      "learning_rate": 9.743209567985043e-05,
      "loss": 1.0779,
      "step": 7255
    },
    {
      "epoch": 0.5230303467166438,
      "grad_norm": 0.1222779451928603,
      "learning_rate": 9.740875629791759e-05,
      "loss": 0.9833,
      "step": 7256
    },
    {
      "epoch": 0.5231024291789808,
      "grad_norm": 0.12374959751753045,
      "learning_rate": 9.738541705723069e-05,
      "loss": 1.0976,
      "step": 7257
    },
    {
      "epoch": 0.5231745116413177,
      "grad_norm": 0.11389248573333571,
      "learning_rate": 9.736207795906194e-05,
      "loss": 1.0504,
      "step": 7258
    },
    {
      "epoch": 0.5232465941036546,
      "grad_norm": 0.13500171845288692,
      "learning_rate": 9.733873900468355e-05,
      "loss": 1.1145,
      "step": 7259
    },
    {
      "epoch": 0.5233186765659915,
      "grad_norm": 0.3183565503154429,
      "learning_rate": 9.731540019536772e-05,
      "loss": 1.0653,
      "step": 7260
    },
    {
      "epoch": 0.5233907590283284,
      "grad_norm": 0.26642610990200977,
      "learning_rate": 9.729206153238657e-05,
      "loss": 1.0317,
      "step": 7261
    },
    {
      "epoch": 0.5234628414906654,
      "grad_norm": 0.12871843909430022,
      "learning_rate": 9.726872301701227e-05,
      "loss": 1.0417,
      "step": 7262
    },
    {
      "epoch": 0.5235349239530023,
      "grad_norm": 0.13078109459377715,
      "learning_rate": 9.724538465051704e-05,
      "loss": 0.938,
      "step": 7263
    },
    {
      "epoch": 0.5236070064153392,
      "grad_norm": 0.13099404086122812,
      "learning_rate": 9.722204643417299e-05,
      "loss": 1.033,
      "step": 7264
    },
    {
      "epoch": 0.5236790888776761,
      "grad_norm": 0.1449726900590414,
      "learning_rate": 9.719870836925224e-05,
      "loss": 1.0186,
      "step": 7265
    },
    {
      "epoch": 0.5237511713400129,
      "grad_norm": 0.1413441388828737,
      "learning_rate": 9.717537045702698e-05,
      "loss": 0.9766,
      "step": 7266
    },
    {
      "epoch": 0.5238232538023498,
      "grad_norm": 0.14816408167625028,
      "learning_rate": 9.715203269876924e-05,
      "loss": 1.0569,
      "step": 7267
    },
    {
      "epoch": 0.5238953362646868,
      "grad_norm": 0.12229430116028926,
      "learning_rate": 9.712869509575124e-05,
      "loss": 1.0858,
      "step": 7268
    },
    {
      "epoch": 0.5239674187270237,
      "grad_norm": 0.12397161304937153,
      "learning_rate": 9.710535764924504e-05,
      "loss": 1.029,
      "step": 7269
    },
    {
      "epoch": 0.5240395011893606,
      "grad_norm": 0.10418245418667833,
      "learning_rate": 9.708202036052276e-05,
      "loss": 1.0198,
      "step": 7270
    },
    {
      "epoch": 0.5241115836516975,
      "grad_norm": 0.18331174867301747,
      "learning_rate": 9.70586832308564e-05,
      "loss": 1.0299,
      "step": 7271
    },
    {
      "epoch": 0.5241836661140344,
      "grad_norm": 0.1334219959514327,
      "learning_rate": 9.70353462615182e-05,
      "loss": 1.041,
      "step": 7272
    },
    {
      "epoch": 0.5242557485763714,
      "grad_norm": 0.1780415659461916,
      "learning_rate": 9.701200945378012e-05,
      "loss": 1.0977,
      "step": 7273
    },
    {
      "epoch": 0.5243278310387083,
      "grad_norm": 0.13971715896045667,
      "learning_rate": 9.698867280891428e-05,
      "loss": 1.05,
      "step": 7274
    },
    {
      "epoch": 0.5243999135010452,
      "grad_norm": 0.11884324196755944,
      "learning_rate": 9.696533632819266e-05,
      "loss": 1.0379,
      "step": 7275
    },
    {
      "epoch": 0.5244719959633821,
      "grad_norm": 0.114797480735904,
      "learning_rate": 9.69420000128874e-05,
      "loss": 1.0869,
      "step": 7276
    },
    {
      "epoch": 0.524544078425719,
      "grad_norm": 0.12416663292994773,
      "learning_rate": 9.69186638642705e-05,
      "loss": 1.1436,
      "step": 7277
    },
    {
      "epoch": 0.524616160888056,
      "grad_norm": 0.12502303046964627,
      "learning_rate": 9.689532788361399e-05,
      "loss": 1.0386,
      "step": 7278
    },
    {
      "epoch": 0.5246882433503929,
      "grad_norm": 0.15920096754300925,
      "learning_rate": 9.687199207218986e-05,
      "loss": 1.0613,
      "step": 7279
    },
    {
      "epoch": 0.5247603258127298,
      "grad_norm": 0.15536261493373688,
      "learning_rate": 9.684865643127013e-05,
      "loss": 1.0508,
      "step": 7280
    },
    {
      "epoch": 0.5248324082750667,
      "grad_norm": 0.18375029269017795,
      "learning_rate": 9.682532096212685e-05,
      "loss": 1.0428,
      "step": 7281
    },
    {
      "epoch": 0.5249044907374036,
      "grad_norm": 0.1203614908993253,
      "learning_rate": 9.680198566603199e-05,
      "loss": 0.999,
      "step": 7282
    },
    {
      "epoch": 0.5249765731997404,
      "grad_norm": 0.19581321307480465,
      "learning_rate": 9.677865054425752e-05,
      "loss": 1.0879,
      "step": 7283
    },
    {
      "epoch": 0.5250486556620774,
      "grad_norm": 0.24812820172500782,
      "learning_rate": 9.675531559807536e-05,
      "loss": 1.0919,
      "step": 7284
    },
    {
      "epoch": 0.5251207381244143,
      "grad_norm": 0.12244620393279793,
      "learning_rate": 9.673198082875756e-05,
      "loss": 1.1141,
      "step": 7285
    },
    {
      "epoch": 0.5251928205867512,
      "grad_norm": 0.17278408339900478,
      "learning_rate": 9.670864623757608e-05,
      "loss": 0.9914,
      "step": 7286
    },
    {
      "epoch": 0.5252649030490881,
      "grad_norm": 0.12908387904692692,
      "learning_rate": 9.668531182580278e-05,
      "loss": 1.0387,
      "step": 7287
    },
    {
      "epoch": 0.525336985511425,
      "grad_norm": 0.21188562228146662,
      "learning_rate": 9.666197759470965e-05,
      "loss": 1.0293,
      "step": 7288
    },
    {
      "epoch": 0.525409067973762,
      "grad_norm": 0.10501642821059601,
      "learning_rate": 9.663864354556858e-05,
      "loss": 1.0956,
      "step": 7289
    },
    {
      "epoch": 0.5254811504360989,
      "grad_norm": 0.14787400555787303,
      "learning_rate": 9.661530967965156e-05,
      "loss": 1.0886,
      "step": 7290
    },
    {
      "epoch": 0.5255532328984358,
      "grad_norm": 0.12906744679982668,
      "learning_rate": 9.65919759982304e-05,
      "loss": 1.051,
      "step": 7291
    },
    {
      "epoch": 0.5256253153607727,
      "grad_norm": 0.11591093122244933,
      "learning_rate": 9.656864250257706e-05,
      "loss": 1.0494,
      "step": 7292
    },
    {
      "epoch": 0.5256973978231096,
      "grad_norm": 0.3688954041148221,
      "learning_rate": 9.654530919396338e-05,
      "loss": 1.0276,
      "step": 7293
    },
    {
      "epoch": 0.5257694802854466,
      "grad_norm": 0.5185464652103388,
      "learning_rate": 9.652197607366129e-05,
      "loss": 1.1141,
      "step": 7294
    },
    {
      "epoch": 0.5258415627477835,
      "grad_norm": 0.10782208145316104,
      "learning_rate": 9.64986431429426e-05,
      "loss": 1.0838,
      "step": 7295
    },
    {
      "epoch": 0.5259136452101204,
      "grad_norm": 0.1317526904049967,
      "learning_rate": 9.647531040307918e-05,
      "loss": 1.046,
      "step": 7296
    },
    {
      "epoch": 0.5259857276724573,
      "grad_norm": 0.13884406370901314,
      "learning_rate": 9.645197785534286e-05,
      "loss": 1.0647,
      "step": 7297
    },
    {
      "epoch": 0.5260578101347942,
      "grad_norm": 0.11036411002137388,
      "learning_rate": 9.642864550100552e-05,
      "loss": 1.0435,
      "step": 7298
    },
    {
      "epoch": 0.5261298925971312,
      "grad_norm": 0.1523224294305331,
      "learning_rate": 9.640531334133894e-05,
      "loss": 1.0087,
      "step": 7299
    },
    {
      "epoch": 0.5262019750594681,
      "grad_norm": 0.11008030519233802,
      "learning_rate": 9.638198137761494e-05,
      "loss": 0.9489,
      "step": 7300
    },
    {
      "epoch": 0.5262740575218049,
      "grad_norm": 0.12583532188014612,
      "learning_rate": 9.635864961110535e-05,
      "loss": 1.0995,
      "step": 7301
    },
    {
      "epoch": 0.5263461399841418,
      "grad_norm": 0.1455738767271455,
      "learning_rate": 9.633531804308187e-05,
      "loss": 1.0262,
      "step": 7302
    },
    {
      "epoch": 0.5264182224464787,
      "grad_norm": 1.5994706315063514,
      "learning_rate": 9.631198667481637e-05,
      "loss": 1.0584,
      "step": 7303
    },
    {
      "epoch": 0.5264903049088157,
      "grad_norm": 0.13472648370508775,
      "learning_rate": 9.62886555075806e-05,
      "loss": 1.0321,
      "step": 7304
    },
    {
      "epoch": 0.5265623873711526,
      "grad_norm": 0.13553026381389485,
      "learning_rate": 9.626532454264633e-05,
      "loss": 1.0668,
      "step": 7305
    },
    {
      "epoch": 0.5266344698334895,
      "grad_norm": 0.12663714781127278,
      "learning_rate": 9.624199378128524e-05,
      "loss": 1.0173,
      "step": 7306
    },
    {
      "epoch": 0.5267065522958264,
      "grad_norm": 0.160174123804891,
      "learning_rate": 9.621866322476911e-05,
      "loss": 1.0407,
      "step": 7307
    },
    {
      "epoch": 0.5267786347581633,
      "grad_norm": 0.12427537938168841,
      "learning_rate": 9.619533287436968e-05,
      "loss": 1.0485,
      "step": 7308
    },
    {
      "epoch": 0.5268507172205003,
      "grad_norm": 0.12286036112133562,
      "learning_rate": 9.617200273135866e-05,
      "loss": 1.0316,
      "step": 7309
    },
    {
      "epoch": 0.5269227996828372,
      "grad_norm": 0.10748819535605943,
      "learning_rate": 9.614867279700774e-05,
      "loss": 1.099,
      "step": 7310
    },
    {
      "epoch": 0.5269948821451741,
      "grad_norm": 0.12194745512321178,
      "learning_rate": 9.612534307258856e-05,
      "loss": 0.959,
      "step": 7311
    },
    {
      "epoch": 0.527066964607511,
      "grad_norm": 0.2046264057101691,
      "learning_rate": 9.610201355937287e-05,
      "loss": 0.9987,
      "step": 7312
    },
    {
      "epoch": 0.5271390470698479,
      "grad_norm": 0.11314253191088992,
      "learning_rate": 9.607868425863235e-05,
      "loss": 1.0217,
      "step": 7313
    },
    {
      "epoch": 0.5272111295321849,
      "grad_norm": 0.1534916328183913,
      "learning_rate": 9.605535517163859e-05,
      "loss": 1.0397,
      "step": 7314
    },
    {
      "epoch": 0.5272832119945218,
      "grad_norm": 0.302962708039061,
      "learning_rate": 9.603202629966323e-05,
      "loss": 1.0601,
      "step": 7315
    },
    {
      "epoch": 0.5273552944568587,
      "grad_norm": 0.10571235257968409,
      "learning_rate": 9.600869764397798e-05,
      "loss": 1.0947,
      "step": 7316
    },
    {
      "epoch": 0.5274273769191956,
      "grad_norm": 0.15782314803225156,
      "learning_rate": 9.598536920585442e-05,
      "loss": 0.9907,
      "step": 7317
    },
    {
      "epoch": 0.5274994593815324,
      "grad_norm": 0.121493296232748,
      "learning_rate": 9.596204098656413e-05,
      "loss": 1.0574,
      "step": 7318
    },
    {
      "epoch": 0.5275715418438693,
      "grad_norm": 0.13139505085935715,
      "learning_rate": 9.59387129873787e-05,
      "loss": 1.1133,
      "step": 7319
    },
    {
      "epoch": 0.5276436243062063,
      "grad_norm": 0.12114407419002822,
      "learning_rate": 9.59153852095698e-05,
      "loss": 0.9713,
      "step": 7320
    },
    {
      "epoch": 0.5277157067685432,
      "grad_norm": 0.11529078260136473,
      "learning_rate": 9.589205765440895e-05,
      "loss": 1.0701,
      "step": 7321
    },
    {
      "epoch": 0.5277877892308801,
      "grad_norm": 0.11571052708266337,
      "learning_rate": 9.586873032316769e-05,
      "loss": 1.1019,
      "step": 7322
    },
    {
      "epoch": 0.527859871693217,
      "grad_norm": 0.11756888729846247,
      "learning_rate": 9.58454032171176e-05,
      "loss": 1.1032,
      "step": 7323
    },
    {
      "epoch": 0.5279319541555539,
      "grad_norm": 0.13062692665541106,
      "learning_rate": 9.582207633753015e-05,
      "loss": 1.0375,
      "step": 7324
    },
    {
      "epoch": 0.5280040366178909,
      "grad_norm": 0.12136113434195368,
      "learning_rate": 9.579874968567697e-05,
      "loss": 1.0889,
      "step": 7325
    },
    {
      "epoch": 0.5280761190802278,
      "grad_norm": 0.1222634591198493,
      "learning_rate": 9.577542326282951e-05,
      "loss": 1.0086,
      "step": 7326
    },
    {
      "epoch": 0.5281482015425647,
      "grad_norm": 0.11311961248167234,
      "learning_rate": 9.575209707025929e-05,
      "loss": 0.9922,
      "step": 7327
    },
    {
      "epoch": 0.5282202840049016,
      "grad_norm": 0.11753732180151244,
      "learning_rate": 9.572877110923772e-05,
      "loss": 1.0433,
      "step": 7328
    },
    {
      "epoch": 0.5282923664672385,
      "grad_norm": 0.22436931163240828,
      "learning_rate": 9.570544538103639e-05,
      "loss": 1.0511,
      "step": 7329
    },
    {
      "epoch": 0.5283644489295755,
      "grad_norm": 0.12311775306231156,
      "learning_rate": 9.56821198869267e-05,
      "loss": 1.0852,
      "step": 7330
    },
    {
      "epoch": 0.5284365313919124,
      "grad_norm": 0.11865145758206645,
      "learning_rate": 9.565879462818012e-05,
      "loss": 1.0177,
      "step": 7331
    },
    {
      "epoch": 0.5285086138542493,
      "grad_norm": 0.18363897670659873,
      "learning_rate": 9.563546960606805e-05,
      "loss": 0.9998,
      "step": 7332
    },
    {
      "epoch": 0.5285806963165862,
      "grad_norm": 0.16670857257622818,
      "learning_rate": 9.561214482186192e-05,
      "loss": 1.111,
      "step": 7333
    },
    {
      "epoch": 0.5286527787789231,
      "grad_norm": 0.1397721823000443,
      "learning_rate": 9.558882027683316e-05,
      "loss": 1.0649,
      "step": 7334
    },
    {
      "epoch": 0.52872486124126,
      "grad_norm": 0.10966401869072277,
      "learning_rate": 9.556549597225319e-05,
      "loss": 1.0351,
      "step": 7335
    },
    {
      "epoch": 0.5287969437035969,
      "grad_norm": 0.13400576232529166,
      "learning_rate": 9.554217190939332e-05,
      "loss": 1.0066,
      "step": 7336
    },
    {
      "epoch": 0.5288690261659338,
      "grad_norm": 0.10220001168678856,
      "learning_rate": 9.551884808952496e-05,
      "loss": 1.0445,
      "step": 7337
    },
    {
      "epoch": 0.5289411086282707,
      "grad_norm": 0.12710323972580398,
      "learning_rate": 9.549552451391947e-05,
      "loss": 1.0127,
      "step": 7338
    },
    {
      "epoch": 0.5290131910906076,
      "grad_norm": 0.13449252396919104,
      "learning_rate": 9.547220118384824e-05,
      "loss": 0.9938,
      "step": 7339
    },
    {
      "epoch": 0.5290852735529445,
      "grad_norm": 0.1267060703084267,
      "learning_rate": 9.544887810058252e-05,
      "loss": 1.04,
      "step": 7340
    },
    {
      "epoch": 0.5291573560152815,
      "grad_norm": 0.11383641274908488,
      "learning_rate": 9.542555526539364e-05,
      "loss": 0.9994,
      "step": 7341
    },
    {
      "epoch": 0.5292294384776184,
      "grad_norm": 0.1436744198901587,
      "learning_rate": 9.540223267955294e-05,
      "loss": 1.0931,
      "step": 7342
    },
    {
      "epoch": 0.5293015209399553,
      "grad_norm": 0.12331995550528958,
      "learning_rate": 9.53789103443317e-05,
      "loss": 1.0633,
      "step": 7343
    },
    {
      "epoch": 0.5293736034022922,
      "grad_norm": 0.14140901393114255,
      "learning_rate": 9.535558826100119e-05,
      "loss": 1.0537,
      "step": 7344
    },
    {
      "epoch": 0.5294456858646291,
      "grad_norm": 0.1276758374573858,
      "learning_rate": 9.533226643083268e-05,
      "loss": 1.1549,
      "step": 7345
    },
    {
      "epoch": 0.5295177683269661,
      "grad_norm": 0.12027448072287142,
      "learning_rate": 9.530894485509735e-05,
      "loss": 1.0542,
      "step": 7346
    },
    {
      "epoch": 0.529589850789303,
      "grad_norm": 0.12121419834330104,
      "learning_rate": 9.528562353506655e-05,
      "loss": 1.0541,
      "step": 7347
    },
    {
      "epoch": 0.5296619332516399,
      "grad_norm": 0.16467474329060253,
      "learning_rate": 9.526230247201142e-05,
      "loss": 1.0408,
      "step": 7348
    },
    {
      "epoch": 0.5297340157139768,
      "grad_norm": 0.2824638449508909,
      "learning_rate": 9.523898166720323e-05,
      "loss": 1.0125,
      "step": 7349
    },
    {
      "epoch": 0.5298060981763137,
      "grad_norm": 0.1309185914037626,
      "learning_rate": 9.521566112191305e-05,
      "loss": 1.0637,
      "step": 7350
    },
    {
      "epoch": 0.5298781806386507,
      "grad_norm": 0.1442099963762339,
      "learning_rate": 9.519234083741221e-05,
      "loss": 1.0886,
      "step": 7351
    },
    {
      "epoch": 0.5299502631009876,
      "grad_norm": 0.11913359561989575,
      "learning_rate": 9.516902081497177e-05,
      "loss": 1.0987,
      "step": 7352
    },
    {
      "epoch": 0.5300223455633244,
      "grad_norm": 0.10776003324570603,
      "learning_rate": 9.514570105586296e-05,
      "loss": 1.0118,
      "step": 7353
    },
    {
      "epoch": 0.5300944280256613,
      "grad_norm": 0.12630574584506668,
      "learning_rate": 9.512238156135684e-05,
      "loss": 0.9982,
      "step": 7354
    },
    {
      "epoch": 0.5301665104879982,
      "grad_norm": 0.12794393065136148,
      "learning_rate": 9.509906233272453e-05,
      "loss": 1.0535,
      "step": 7355
    },
    {
      "epoch": 0.5302385929503352,
      "grad_norm": 0.20068849707312789,
      "learning_rate": 9.507574337123719e-05,
      "loss": 1.06,
      "step": 7356
    },
    {
      "epoch": 0.5303106754126721,
      "grad_norm": 0.11452144229866101,
      "learning_rate": 9.505242467816592e-05,
      "loss": 1.0099,
      "step": 7357
    },
    {
      "epoch": 0.530382757875009,
      "grad_norm": 0.14274405401085524,
      "learning_rate": 9.502910625478175e-05,
      "loss": 1.095,
      "step": 7358
    },
    {
      "epoch": 0.5304548403373459,
      "grad_norm": 0.12422443801098661,
      "learning_rate": 9.500578810235573e-05,
      "loss": 0.9847,
      "step": 7359
    },
    {
      "epoch": 0.5305269227996828,
      "grad_norm": 0.11794199716460038,
      "learning_rate": 9.498247022215896e-05,
      "loss": 1.0438,
      "step": 7360
    },
    {
      "epoch": 0.5305990052620198,
      "grad_norm": 0.1256315455167962,
      "learning_rate": 9.495915261546247e-05,
      "loss": 1.0664,
      "step": 7361
    },
    {
      "epoch": 0.5306710877243567,
      "grad_norm": 0.13774649954056986,
      "learning_rate": 9.493583528353724e-05,
      "loss": 1.029,
      "step": 7362
    },
    {
      "epoch": 0.5307431701866936,
      "grad_norm": 0.16214006421810565,
      "learning_rate": 9.491251822765427e-05,
      "loss": 1.1112,
      "step": 7363
    },
    {
      "epoch": 0.5308152526490305,
      "grad_norm": 0.14661029672308126,
      "learning_rate": 9.48892014490846e-05,
      "loss": 1.0242,
      "step": 7364
    },
    {
      "epoch": 0.5308873351113674,
      "grad_norm": 0.19066722996095478,
      "learning_rate": 9.48658849490992e-05,
      "loss": 1.0768,
      "step": 7365
    },
    {
      "epoch": 0.5309594175737044,
      "grad_norm": 0.11201374759001075,
      "learning_rate": 9.484256872896896e-05,
      "loss": 1.0258,
      "step": 7366
    },
    {
      "epoch": 0.5310315000360413,
      "grad_norm": 0.13825065884461146,
      "learning_rate": 9.481925278996488e-05,
      "loss": 1.0496,
      "step": 7367
    },
    {
      "epoch": 0.5311035824983782,
      "grad_norm": 0.12582930393298336,
      "learning_rate": 9.479593713335784e-05,
      "loss": 1.0687,
      "step": 7368
    },
    {
      "epoch": 0.5311756649607151,
      "grad_norm": 0.150311162576532,
      "learning_rate": 9.477262176041883e-05,
      "loss": 1.1687,
      "step": 7369
    },
    {
      "epoch": 0.5312477474230519,
      "grad_norm": 0.11302681880563645,
      "learning_rate": 9.474930667241868e-05,
      "loss": 1.0637,
      "step": 7370
    },
    {
      "epoch": 0.5313198298853888,
      "grad_norm": 0.18170774640163606,
      "learning_rate": 9.472599187062829e-05,
      "loss": 1.0657,
      "step": 7371
    },
    {
      "epoch": 0.5313919123477258,
      "grad_norm": 0.1265609871641912,
      "learning_rate": 9.47026773563185e-05,
      "loss": 0.95,
      "step": 7372
    },
    {
      "epoch": 0.5314639948100627,
      "grad_norm": 0.11181506488580475,
      "learning_rate": 9.467936313076022e-05,
      "loss": 1.1017,
      "step": 7373
    },
    {
      "epoch": 0.5315360772723996,
      "grad_norm": 0.11221580034400944,
      "learning_rate": 9.465604919522425e-05,
      "loss": 1.1115,
      "step": 7374
    },
    {
      "epoch": 0.5316081597347365,
      "grad_norm": 0.14426218860840176,
      "learning_rate": 9.46327355509814e-05,
      "loss": 1.0613,
      "step": 7375
    },
    {
      "epoch": 0.5316802421970734,
      "grad_norm": 0.34935807840047184,
      "learning_rate": 9.460942219930248e-05,
      "loss": 1.0281,
      "step": 7376
    },
    {
      "epoch": 0.5317523246594104,
      "grad_norm": 0.14017535712137505,
      "learning_rate": 9.458610914145826e-05,
      "loss": 0.9803,
      "step": 7377
    },
    {
      "epoch": 0.5318244071217473,
      "grad_norm": 0.11111482453214623,
      "learning_rate": 9.456279637871953e-05,
      "loss": 1.1285,
      "step": 7378
    },
    {
      "epoch": 0.5318964895840842,
      "grad_norm": 0.11589622337608911,
      "learning_rate": 9.453948391235706e-05,
      "loss": 0.986,
      "step": 7379
    },
    {
      "epoch": 0.5319685720464211,
      "grad_norm": 0.19478886581232147,
      "learning_rate": 9.451617174364157e-05,
      "loss": 1.0641,
      "step": 7380
    },
    {
      "epoch": 0.532040654508758,
      "grad_norm": 0.5095849389196305,
      "learning_rate": 9.449285987384373e-05,
      "loss": 1.0594,
      "step": 7381
    },
    {
      "epoch": 0.532112736971095,
      "grad_norm": 0.13239216971402,
      "learning_rate": 9.446954830423434e-05,
      "loss": 1.0295,
      "step": 7382
    },
    {
      "epoch": 0.5321848194334319,
      "grad_norm": 0.13914473278417652,
      "learning_rate": 9.444623703608402e-05,
      "loss": 1.0771,
      "step": 7383
    },
    {
      "epoch": 0.5322569018957688,
      "grad_norm": 0.13367017201520337,
      "learning_rate": 9.442292607066351e-05,
      "loss": 1.0326,
      "step": 7384
    },
    {
      "epoch": 0.5323289843581057,
      "grad_norm": 0.11958579503083368,
      "learning_rate": 9.439961540924337e-05,
      "loss": 1.0553,
      "step": 7385
    },
    {
      "epoch": 0.5324010668204426,
      "grad_norm": 0.141491349284102,
      "learning_rate": 9.437630505309432e-05,
      "loss": 1.0186,
      "step": 7386
    },
    {
      "epoch": 0.5324731492827794,
      "grad_norm": 0.4931941288596791,
      "learning_rate": 9.435299500348695e-05,
      "loss": 1.0091,
      "step": 7387
    },
    {
      "epoch": 0.5325452317451164,
      "grad_norm": 0.10351797928324227,
      "learning_rate": 9.43296852616919e-05,
      "loss": 1.0451,
      "step": 7388
    },
    {
      "epoch": 0.5326173142074533,
      "grad_norm": 0.11437950939474811,
      "learning_rate": 9.43063758289797e-05,
      "loss": 1.0428,
      "step": 7389
    },
    {
      "epoch": 0.5326893966697902,
      "grad_norm": 0.13554479740253633,
      "learning_rate": 9.428306670662093e-05,
      "loss": 1.0247,
      "step": 7390
    },
    {
      "epoch": 0.5327614791321271,
      "grad_norm": 0.14473213889659786,
      "learning_rate": 9.425975789588621e-05,
      "loss": 0.9711,
      "step": 7391
    },
    {
      "epoch": 0.532833561594464,
      "grad_norm": 0.2720017585712402,
      "learning_rate": 9.423644939804605e-05,
      "loss": 1.028,
      "step": 7392
    },
    {
      "epoch": 0.532905644056801,
      "grad_norm": 0.13012321011530822,
      "learning_rate": 9.421314121437093e-05,
      "loss": 1.044,
      "step": 7393
    },
    {
      "epoch": 0.5329777265191379,
      "grad_norm": 0.15691279011358866,
      "learning_rate": 9.418983334613136e-05,
      "loss": 1.092,
      "step": 7394
    },
    {
      "epoch": 0.5330498089814748,
      "grad_norm": 0.15124457538711872,
      "learning_rate": 9.416652579459789e-05,
      "loss": 1.1227,
      "step": 7395
    },
    {
      "epoch": 0.5331218914438117,
      "grad_norm": 0.1261524355754968,
      "learning_rate": 9.414321856104097e-05,
      "loss": 0.9828,
      "step": 7396
    },
    {
      "epoch": 0.5331939739061486,
      "grad_norm": 0.1280412172877746,
      "learning_rate": 9.4119911646731e-05,
      "loss": 0.9683,
      "step": 7397
    },
    {
      "epoch": 0.5332660563684856,
      "grad_norm": 0.15147785410256587,
      "learning_rate": 9.409660505293849e-05,
      "loss": 1.0637,
      "step": 7398
    },
    {
      "epoch": 0.5333381388308225,
      "grad_norm": 0.13579562511643162,
      "learning_rate": 9.407329878093374e-05,
      "loss": 1.0821,
      "step": 7399
    },
    {
      "epoch": 0.5334102212931594,
      "grad_norm": 0.1086108454179448,
      "learning_rate": 9.40499928319873e-05,
      "loss": 1.0695,
      "step": 7400
    },
    {
      "epoch": 0.5334823037554963,
      "grad_norm": 0.4327234000827465,
      "learning_rate": 9.402668720736946e-05,
      "loss": 1.112,
      "step": 7401
    },
    {
      "epoch": 0.5335543862178332,
      "grad_norm": 0.11171280907227023,
      "learning_rate": 9.400338190835062e-05,
      "loss": 1.047,
      "step": 7402
    },
    {
      "epoch": 0.5336264686801702,
      "grad_norm": 0.20022944874349938,
      "learning_rate": 9.398007693620105e-05,
      "loss": 1.1142,
      "step": 7403
    },
    {
      "epoch": 0.5336985511425071,
      "grad_norm": 0.12101461927048975,
      "learning_rate": 9.395677229219121e-05,
      "loss": 1.0562,
      "step": 7404
    },
    {
      "epoch": 0.5337706336048439,
      "grad_norm": 0.1357091058758094,
      "learning_rate": 9.393346797759133e-05,
      "loss": 0.9921,
      "step": 7405
    },
    {
      "epoch": 0.5338427160671808,
      "grad_norm": 0.24244180066181176,
      "learning_rate": 9.391016399367172e-05,
      "loss": 0.925,
      "step": 7406
    },
    {
      "epoch": 0.5339147985295177,
      "grad_norm": 0.12359388561583082,
      "learning_rate": 9.388686034170261e-05,
      "loss": 1.0542,
      "step": 7407
    },
    {
      "epoch": 0.5339868809918547,
      "grad_norm": 0.13128817498030076,
      "learning_rate": 9.386355702295435e-05,
      "loss": 1.0313,
      "step": 7408
    },
    {
      "epoch": 0.5340589634541916,
      "grad_norm": 0.14925883170238333,
      "learning_rate": 9.384025403869711e-05,
      "loss": 1.0838,
      "step": 7409
    },
    {
      "epoch": 0.5341310459165285,
      "grad_norm": 0.12405528584084935,
      "learning_rate": 9.381695139020117e-05,
      "loss": 1.0072,
      "step": 7410
    },
    {
      "epoch": 0.5342031283788654,
      "grad_norm": 0.13871673692514236,
      "learning_rate": 9.379364907873667e-05,
      "loss": 0.9733,
      "step": 7411
    },
    {
      "epoch": 0.5342752108412023,
      "grad_norm": 0.1454219522943042,
      "learning_rate": 9.377034710557378e-05,
      "loss": 1.0813,
      "step": 7412
    },
    {
      "epoch": 0.5343472933035393,
      "grad_norm": 0.11587212916286521,
      "learning_rate": 9.374704547198274e-05,
      "loss": 1.0174,
      "step": 7413
    },
    {
      "epoch": 0.5344193757658762,
      "grad_norm": 0.1195812610620097,
      "learning_rate": 9.372374417923367e-05,
      "loss": 1.0523,
      "step": 7414
    },
    {
      "epoch": 0.5344914582282131,
      "grad_norm": 0.12250585719992431,
      "learning_rate": 9.370044322859669e-05,
      "loss": 1.1033,
      "step": 7415
    },
    {
      "epoch": 0.53456354069055,
      "grad_norm": 0.11655008995565211,
      "learning_rate": 9.367714262134188e-05,
      "loss": 1.1057,
      "step": 7416
    },
    {
      "epoch": 0.5346356231528869,
      "grad_norm": 0.11801279395704849,
      "learning_rate": 9.365384235873938e-05,
      "loss": 0.9794,
      "step": 7417
    },
    {
      "epoch": 0.5347077056152238,
      "grad_norm": 1.3041656366640004,
      "learning_rate": 9.363054244205927e-05,
      "loss": 1.0725,
      "step": 7418
    },
    {
      "epoch": 0.5347797880775608,
      "grad_norm": 0.12048336459864946,
      "learning_rate": 9.360724287257156e-05,
      "loss": 1.081,
      "step": 7419
    },
    {
      "epoch": 0.5348518705398977,
      "grad_norm": 0.17366114855336887,
      "learning_rate": 9.358394365154632e-05,
      "loss": 1.0767,
      "step": 7420
    },
    {
      "epoch": 0.5349239530022346,
      "grad_norm": 1.1739849036724974,
      "learning_rate": 9.356064478025349e-05,
      "loss": 1.1307,
      "step": 7421
    },
    {
      "epoch": 0.5349960354645714,
      "grad_norm": 0.1063250112637127,
      "learning_rate": 9.35373462599632e-05,
      "loss": 1.0059,
      "step": 7422
    },
    {
      "epoch": 0.5350681179269083,
      "grad_norm": 2.2275449974082524,
      "learning_rate": 9.351404809194533e-05,
      "loss": 1.0553,
      "step": 7423
    },
    {
      "epoch": 0.5351402003892453,
      "grad_norm": 0.11094138380483437,
      "learning_rate": 9.349075027746987e-05,
      "loss": 1.055,
      "step": 7424
    },
    {
      "epoch": 0.5352122828515822,
      "grad_norm": 0.10721606325598883,
      "learning_rate": 9.346745281780671e-05,
      "loss": 1.0494,
      "step": 7425
    },
    {
      "epoch": 0.5352843653139191,
      "grad_norm": 0.18462134248766265,
      "learning_rate": 9.344415571422586e-05,
      "loss": 1.0588,
      "step": 7426
    },
    {
      "epoch": 0.535356447776256,
      "grad_norm": 0.14045270259013,
      "learning_rate": 9.342085896799718e-05,
      "loss": 1.0071,
      "step": 7427
    },
    {
      "epoch": 0.5354285302385929,
      "grad_norm": 0.11688040545760936,
      "learning_rate": 9.339756258039054e-05,
      "loss": 1.1121,
      "step": 7428
    },
    {
      "epoch": 0.5355006127009299,
      "grad_norm": 0.1152821865713924,
      "learning_rate": 9.33742665526758e-05,
      "loss": 1.0259,
      "step": 7429
    },
    {
      "epoch": 0.5355726951632668,
      "grad_norm": 0.13191061262831105,
      "learning_rate": 9.335097088612276e-05,
      "loss": 1.0656,
      "step": 7430
    },
    {
      "epoch": 0.5356447776256037,
      "grad_norm": 0.11389106861034493,
      "learning_rate": 9.332767558200135e-05,
      "loss": 0.9942,
      "step": 7431
    },
    {
      "epoch": 0.5357168600879406,
      "grad_norm": 0.10995514012629487,
      "learning_rate": 9.330438064158131e-05,
      "loss": 1.0279,
      "step": 7432
    },
    {
      "epoch": 0.5357889425502775,
      "grad_norm": 0.13113294073484855,
      "learning_rate": 9.328108606613241e-05,
      "loss": 1.0951,
      "step": 7433
    },
    {
      "epoch": 0.5358610250126145,
      "grad_norm": 0.12428249205759107,
      "learning_rate": 9.325779185692441e-05,
      "loss": 1.0523,
      "step": 7434
    },
    {
      "epoch": 0.5359331074749514,
      "grad_norm": 0.21754781638805137,
      "learning_rate": 9.323449801522709e-05,
      "loss": 1.007,
      "step": 7435
    },
    {
      "epoch": 0.5360051899372883,
      "grad_norm": 0.15799327187163703,
      "learning_rate": 9.321120454231019e-05,
      "loss": 1.0528,
      "step": 7436
    },
    {
      "epoch": 0.5360772723996252,
      "grad_norm": 0.12612076959635973,
      "learning_rate": 9.318791143944333e-05,
      "loss": 1.1132,
      "step": 7437
    },
    {
      "epoch": 0.5361493548619621,
      "grad_norm": 0.10914030867251454,
      "learning_rate": 9.316461870789624e-05,
      "loss": 1.0033,
      "step": 7438
    },
    {
      "epoch": 0.5362214373242989,
      "grad_norm": 0.14414183507297962,
      "learning_rate": 9.314132634893859e-05,
      "loss": 1.0576,
      "step": 7439
    },
    {
      "epoch": 0.5362935197866359,
      "grad_norm": 0.11978079602590987,
      "learning_rate": 9.311803436384003e-05,
      "loss": 0.993,
      "step": 7440
    },
    {
      "epoch": 0.5363656022489728,
      "grad_norm": 0.1250293611757527,
      "learning_rate": 9.309474275387015e-05,
      "loss": 1.0378,
      "step": 7441
    },
    {
      "epoch": 0.5364376847113097,
      "grad_norm": 0.13186602183063934,
      "learning_rate": 9.307145152029857e-05,
      "loss": 1.0185,
      "step": 7442
    },
    {
      "epoch": 0.5365097671736466,
      "grad_norm": 0.10994178378461535,
      "learning_rate": 9.304816066439484e-05,
      "loss": 1.0855,
      "step": 7443
    },
    {
      "epoch": 0.5365818496359835,
      "grad_norm": 0.17794725048629348,
      "learning_rate": 9.30248701874286e-05,
      "loss": 0.9926,
      "step": 7444
    },
    {
      "epoch": 0.5366539320983205,
      "grad_norm": 0.5243100007777358,
      "learning_rate": 9.30015800906693e-05,
      "loss": 1.1756,
      "step": 7445
    },
    {
      "epoch": 0.5367260145606574,
      "grad_norm": 1.32250804263455,
      "learning_rate": 9.29782903753865e-05,
      "loss": 1.0103,
      "step": 7446
    },
    {
      "epoch": 0.5367980970229943,
      "grad_norm": 0.23849894860303306,
      "learning_rate": 9.295500104284967e-05,
      "loss": 0.966,
      "step": 7447
    },
    {
      "epoch": 0.5368701794853312,
      "grad_norm": 0.1433066586567518,
      "learning_rate": 9.293171209432836e-05,
      "loss": 1.1296,
      "step": 7448
    },
    {
      "epoch": 0.5369422619476681,
      "grad_norm": 0.10716579167047804,
      "learning_rate": 9.290842353109195e-05,
      "loss": 1.0673,
      "step": 7449
    },
    {
      "epoch": 0.5370143444100051,
      "grad_norm": 0.3338535756849636,
      "learning_rate": 9.288513535440989e-05,
      "loss": 1.1075,
      "step": 7450
    },
    {
      "epoch": 0.537086426872342,
      "grad_norm": 0.1439374516311297,
      "learning_rate": 9.286184756555164e-05,
      "loss": 1.0798,
      "step": 7451
    },
    {
      "epoch": 0.5371585093346789,
      "grad_norm": 0.14507553205555326,
      "learning_rate": 9.283856016578651e-05,
      "loss": 0.9817,
      "step": 7452
    },
    {
      "epoch": 0.5372305917970158,
      "grad_norm": 0.13082486846098407,
      "learning_rate": 9.281527315638394e-05,
      "loss": 1.005,
      "step": 7453
    },
    {
      "epoch": 0.5373026742593527,
      "grad_norm": 0.21103015543051223,
      "learning_rate": 9.279198653861327e-05,
      "loss": 1.0329,
      "step": 7454
    },
    {
      "epoch": 0.5373747567216897,
      "grad_norm": 0.1349725973688542,
      "learning_rate": 9.276870031374382e-05,
      "loss": 1.083,
      "step": 7455
    },
    {
      "epoch": 0.5374468391840266,
      "grad_norm": 0.10871089121406666,
      "learning_rate": 9.274541448304486e-05,
      "loss": 1.0896,
      "step": 7456
    },
    {
      "epoch": 0.5375189216463634,
      "grad_norm": 0.13297424038982886,
      "learning_rate": 9.272212904778575e-05,
      "loss": 1.0612,
      "step": 7457
    },
    {
      "epoch": 0.5375910041087003,
      "grad_norm": 0.15734050842541403,
      "learning_rate": 9.26988440092357e-05,
      "loss": 1.0806,
      "step": 7458
    },
    {
      "epoch": 0.5376630865710372,
      "grad_norm": 0.13686731404508296,
      "learning_rate": 9.2675559368664e-05,
      "loss": 0.9869,
      "step": 7459
    },
    {
      "epoch": 0.5377351690333741,
      "grad_norm": 0.12224813584285138,
      "learning_rate": 9.265227512733978e-05,
      "loss": 1.0625,
      "step": 7460
    },
    {
      "epoch": 0.5378072514957111,
      "grad_norm": 0.11862225794241589,
      "learning_rate": 9.262899128653233e-05,
      "loss": 1.02,
      "step": 7461
    },
    {
      "epoch": 0.537879333958048,
      "grad_norm": 0.110160691324098,
      "learning_rate": 9.26057078475108e-05,
      "loss": 0.9776,
      "step": 7462
    },
    {
      "epoch": 0.5379514164203849,
      "grad_norm": 0.14395771820863404,
      "learning_rate": 9.258242481154437e-05,
      "loss": 1.0226,
      "step": 7463
    },
    {
      "epoch": 0.5380234988827218,
      "grad_norm": 0.2930489657351852,
      "learning_rate": 9.255914217990211e-05,
      "loss": 1.0021,
      "step": 7464
    },
    {
      "epoch": 0.5380955813450587,
      "grad_norm": 0.10798610176445442,
      "learning_rate": 9.253585995385317e-05,
      "loss": 1.1095,
      "step": 7465
    },
    {
      "epoch": 0.5381676638073957,
      "grad_norm": 0.12519311864874974,
      "learning_rate": 9.251257813466664e-05,
      "loss": 1.0798,
      "step": 7466
    },
    {
      "epoch": 0.5382397462697326,
      "grad_norm": 0.6976177176267503,
      "learning_rate": 9.248929672361163e-05,
      "loss": 1.0212,
      "step": 7467
    },
    {
      "epoch": 0.5383118287320695,
      "grad_norm": 0.16188960655166032,
      "learning_rate": 9.24660157219571e-05,
      "loss": 1.0577,
      "step": 7468
    },
    {
      "epoch": 0.5383839111944064,
      "grad_norm": 0.13345423383755878,
      "learning_rate": 9.244273513097211e-05,
      "loss": 1.0543,
      "step": 7469
    },
    {
      "epoch": 0.5384559936567433,
      "grad_norm": 0.14226289349960194,
      "learning_rate": 9.241945495192567e-05,
      "loss": 1.0108,
      "step": 7470
    },
    {
      "epoch": 0.5385280761190803,
      "grad_norm": 0.1286904482057428,
      "learning_rate": 9.239617518608678e-05,
      "loss": 1.0273,
      "step": 7471
    },
    {
      "epoch": 0.5386001585814172,
      "grad_norm": 0.12297310342351346,
      "learning_rate": 9.237289583472434e-05,
      "loss": 1.0353,
      "step": 7472
    },
    {
      "epoch": 0.5386722410437541,
      "grad_norm": 0.1086143185019013,
      "learning_rate": 9.234961689910734e-05,
      "loss": 0.9997,
      "step": 7473
    },
    {
      "epoch": 0.5387443235060909,
      "grad_norm": 0.12008833384561056,
      "learning_rate": 9.232633838050459e-05,
      "loss": 1.0534,
      "step": 7474
    },
    {
      "epoch": 0.5388164059684278,
      "grad_norm": 0.13420131920778594,
      "learning_rate": 9.230306028018513e-05,
      "loss": 1.0246,
      "step": 7475
    },
    {
      "epoch": 0.5388884884307648,
      "grad_norm": 0.11093856422036982,
      "learning_rate": 9.227978259941771e-05,
      "loss": 0.9604,
      "step": 7476
    },
    {
      "epoch": 0.5389605708931017,
      "grad_norm": 0.11639989665734739,
      "learning_rate": 9.225650533947122e-05,
      "loss": 0.9817,
      "step": 7477
    },
    {
      "epoch": 0.5390326533554386,
      "grad_norm": 0.11821234201854611,
      "learning_rate": 9.223322850161441e-05,
      "loss": 1.1421,
      "step": 7478
    },
    {
      "epoch": 0.5391047358177755,
      "grad_norm": 0.1281119349514455,
      "learning_rate": 9.220995208711619e-05,
      "loss": 1.1016,
      "step": 7479
    },
    {
      "epoch": 0.5391768182801124,
      "grad_norm": 0.12009783703806691,
      "learning_rate": 9.218667609724526e-05,
      "loss": 0.9727,
      "step": 7480
    },
    {
      "epoch": 0.5392489007424494,
      "grad_norm": 0.8608137407630211,
      "learning_rate": 9.216340053327039e-05,
      "loss": 0.9815,
      "step": 7481
    },
    {
      "epoch": 0.5393209832047863,
      "grad_norm": 0.1377481926268022,
      "learning_rate": 9.214012539646025e-05,
      "loss": 1.1003,
      "step": 7482
    },
    {
      "epoch": 0.5393930656671232,
      "grad_norm": 0.2223250541425323,
      "learning_rate": 9.211685068808364e-05,
      "loss": 1.0625,
      "step": 7483
    },
    {
      "epoch": 0.5394651481294601,
      "grad_norm": 0.12645179736454218,
      "learning_rate": 9.209357640940919e-05,
      "loss": 1.0069,
      "step": 7484
    },
    {
      "epoch": 0.539537230591797,
      "grad_norm": 0.11422705994762859,
      "learning_rate": 9.207030256170557e-05,
      "loss": 1.0578,
      "step": 7485
    },
    {
      "epoch": 0.539609313054134,
      "grad_norm": 0.11884859320899234,
      "learning_rate": 9.204702914624138e-05,
      "loss": 1.0187,
      "step": 7486
    },
    {
      "epoch": 0.5396813955164709,
      "grad_norm": 0.24736586374913544,
      "learning_rate": 9.202375616428522e-05,
      "loss": 1.0871,
      "step": 7487
    },
    {
      "epoch": 0.5397534779788078,
      "grad_norm": 0.14507163512486712,
      "learning_rate": 9.200048361710575e-05,
      "loss": 0.9414,
      "step": 7488
    },
    {
      "epoch": 0.5398255604411447,
      "grad_norm": 0.13145107974580802,
      "learning_rate": 9.19772115059715e-05,
      "loss": 1.0848,
      "step": 7489
    },
    {
      "epoch": 0.5398976429034816,
      "grad_norm": 0.6335672278173905,
      "learning_rate": 9.195393983215097e-05,
      "loss": 1.1568,
      "step": 7490
    },
    {
      "epoch": 0.5399697253658184,
      "grad_norm": 0.1410467044237045,
      "learning_rate": 9.193066859691268e-05,
      "loss": 0.9962,
      "step": 7491
    },
    {
      "epoch": 0.5400418078281554,
      "grad_norm": 0.6248390764107284,
      "learning_rate": 9.190739780152516e-05,
      "loss": 1.0665,
      "step": 7492
    },
    {
      "epoch": 0.5401138902904923,
      "grad_norm": 0.1796347600430471,
      "learning_rate": 9.18841274472569e-05,
      "loss": 1.0418,
      "step": 7493
    },
    {
      "epoch": 0.5401859727528292,
      "grad_norm": 0.20787482981308908,
      "learning_rate": 9.186085753537625e-05,
      "loss": 0.9661,
      "step": 7494
    },
    {
      "epoch": 0.5402580552151661,
      "grad_norm": 0.11533553167187625,
      "learning_rate": 9.183758806715172e-05,
      "loss": 1.0127,
      "step": 7495
    },
    {
      "epoch": 0.540330137677503,
      "grad_norm": 0.1112921906463289,
      "learning_rate": 9.18143190438516e-05,
      "loss": 1.0487,
      "step": 7496
    },
    {
      "epoch": 0.54040222013984,
      "grad_norm": 0.24268350973290434,
      "learning_rate": 9.179105046674438e-05,
      "loss": 0.9892,
      "step": 7497
    },
    {
      "epoch": 0.5404743026021769,
      "grad_norm": 0.19129464486076214,
      "learning_rate": 9.176778233709833e-05,
      "loss": 1.0935,
      "step": 7498
    },
    {
      "epoch": 0.5405463850645138,
      "grad_norm": 0.12465293401236321,
      "learning_rate": 9.174451465618182e-05,
      "loss": 0.9761,
      "step": 7499
    },
    {
      "epoch": 0.5406184675268507,
      "grad_norm": 0.17029783722888742,
      "learning_rate": 9.172124742526303e-05,
      "loss": 1.11,
      "step": 7500
    },
    {
      "epoch": 0.5406905499891876,
      "grad_norm": 0.1653895223492102,
      "learning_rate": 9.169798064561041e-05,
      "loss": 1.0803,
      "step": 7501
    },
    {
      "epoch": 0.5407626324515246,
      "grad_norm": 0.12965885347877118,
      "learning_rate": 9.167471431849207e-05,
      "loss": 0.9993,
      "step": 7502
    },
    {
      "epoch": 0.5408347149138615,
      "grad_norm": 0.11993575392877387,
      "learning_rate": 9.165144844517632e-05,
      "loss": 1.0845,
      "step": 7503
    },
    {
      "epoch": 0.5409067973761984,
      "grad_norm": 0.17787936322436976,
      "learning_rate": 9.162818302693124e-05,
      "loss": 0.9992,
      "step": 7504
    },
    {
      "epoch": 0.5409788798385353,
      "grad_norm": 0.11769145017319708,
      "learning_rate": 9.160491806502516e-05,
      "loss": 1.0514,
      "step": 7505
    },
    {
      "epoch": 0.5410509623008722,
      "grad_norm": 0.12227693240238535,
      "learning_rate": 9.158165356072612e-05,
      "loss": 1.0264,
      "step": 7506
    },
    {
      "epoch": 0.5411230447632092,
      "grad_norm": 0.23318260288958167,
      "learning_rate": 9.15583895153023e-05,
      "loss": 1.0386,
      "step": 7507
    },
    {
      "epoch": 0.5411951272255461,
      "grad_norm": 0.11968749874564824,
      "learning_rate": 9.153512593002175e-05,
      "loss": 1.0575,
      "step": 7508
    },
    {
      "epoch": 0.5412672096878829,
      "grad_norm": 0.19133974818704017,
      "learning_rate": 9.151186280615253e-05,
      "loss": 1.1021,
      "step": 7509
    },
    {
      "epoch": 0.5413392921502198,
      "grad_norm": 0.108305688400993,
      "learning_rate": 9.148860014496277e-05,
      "loss": 0.9883,
      "step": 7510
    },
    {
      "epoch": 0.5414113746125567,
      "grad_norm": 0.11674173603173293,
      "learning_rate": 9.146533794772046e-05,
      "loss": 1.0468,
      "step": 7511
    },
    {
      "epoch": 0.5414834570748936,
      "grad_norm": 0.17597237828279114,
      "learning_rate": 9.144207621569356e-05,
      "loss": 1.0375,
      "step": 7512
    },
    {
      "epoch": 0.5415555395372306,
      "grad_norm": 0.15213897473291266,
      "learning_rate": 9.141881495015004e-05,
      "loss": 1.0575,
      "step": 7513
    },
    {
      "epoch": 0.5416276219995675,
      "grad_norm": 0.12319543664885191,
      "learning_rate": 9.13955541523579e-05,
      "loss": 1.0378,
      "step": 7514
    },
    {
      "epoch": 0.5416997044619044,
      "grad_norm": 0.14960601155436928,
      "learning_rate": 9.137229382358507e-05,
      "loss": 1.0758,
      "step": 7515
    },
    {
      "epoch": 0.5417717869242413,
      "grad_norm": 0.16270934845645746,
      "learning_rate": 9.134903396509938e-05,
      "loss": 1.0774,
      "step": 7516
    },
    {
      "epoch": 0.5418438693865782,
      "grad_norm": 0.18741920124584416,
      "learning_rate": 9.132577457816873e-05,
      "loss": 1.0491,
      "step": 7517
    },
    {
      "epoch": 0.5419159518489152,
      "grad_norm": 0.14836247618411977,
      "learning_rate": 9.130251566406095e-05,
      "loss": 1.1068,
      "step": 7518
    },
    {
      "epoch": 0.5419880343112521,
      "grad_norm": 0.11470251426364331,
      "learning_rate": 9.127925722404391e-05,
      "loss": 1.0338,
      "step": 7519
    },
    {
      "epoch": 0.542060116773589,
      "grad_norm": 0.11488948531334611,
      "learning_rate": 9.125599925938537e-05,
      "loss": 1.0846,
      "step": 7520
    },
    {
      "epoch": 0.5421321992359259,
      "grad_norm": 0.18421367177958295,
      "learning_rate": 9.123274177135308e-05,
      "loss": 1.007,
      "step": 7521
    },
    {
      "epoch": 0.5422042816982628,
      "grad_norm": 0.11052155692846648,
      "learning_rate": 9.120948476121479e-05,
      "loss": 1.0628,
      "step": 7522
    },
    {
      "epoch": 0.5422763641605998,
      "grad_norm": 0.10243252007453107,
      "learning_rate": 9.118622823023823e-05,
      "loss": 1.0955,
      "step": 7523
    },
    {
      "epoch": 0.5423484466229367,
      "grad_norm": 0.14415199622867753,
      "learning_rate": 9.11629721796911e-05,
      "loss": 1.0965,
      "step": 7524
    },
    {
      "epoch": 0.5424205290852736,
      "grad_norm": 0.1743397415040028,
      "learning_rate": 9.113971661084102e-05,
      "loss": 1.0524,
      "step": 7525
    },
    {
      "epoch": 0.5424926115476104,
      "grad_norm": 0.10409655427182053,
      "learning_rate": 9.111646152495565e-05,
      "loss": 1.0408,
      "step": 7526
    },
    {
      "epoch": 0.5425646940099473,
      "grad_norm": 0.10870635116619749,
      "learning_rate": 9.109320692330263e-05,
      "loss": 0.9943,
      "step": 7527
    },
    {
      "epoch": 0.5426367764722843,
      "grad_norm": 0.13179059658009257,
      "learning_rate": 9.10699528071495e-05,
      "loss": 0.9712,
      "step": 7528
    },
    {
      "epoch": 0.5427088589346212,
      "grad_norm": 0.15088396541535568,
      "learning_rate": 9.104669917776383e-05,
      "loss": 1.0634,
      "step": 7529
    },
    {
      "epoch": 0.5427809413969581,
      "grad_norm": 0.12624995395645297,
      "learning_rate": 9.102344603641318e-05,
      "loss": 0.9972,
      "step": 7530
    },
    {
      "epoch": 0.542853023859295,
      "grad_norm": 0.11388894453937953,
      "learning_rate": 9.100019338436496e-05,
      "loss": 0.9941,
      "step": 7531
    },
    {
      "epoch": 0.5429251063216319,
      "grad_norm": 0.16002034166567589,
      "learning_rate": 9.097694122288676e-05,
      "loss": 1.1103,
      "step": 7532
    },
    {
      "epoch": 0.5429971887839689,
      "grad_norm": 0.09969140388981193,
      "learning_rate": 9.095368955324599e-05,
      "loss": 0.9903,
      "step": 7533
    },
    {
      "epoch": 0.5430692712463058,
      "grad_norm": 0.11686489599796124,
      "learning_rate": 9.093043837671008e-05,
      "loss": 1.0671,
      "step": 7534
    },
    {
      "epoch": 0.5431413537086427,
      "grad_norm": 0.12722968614303837,
      "learning_rate": 9.090718769454636e-05,
      "loss": 1.059,
      "step": 7535
    },
    {
      "epoch": 0.5432134361709796,
      "grad_norm": 0.11448103454958475,
      "learning_rate": 9.088393750802231e-05,
      "loss": 1.0957,
      "step": 7536
    },
    {
      "epoch": 0.5432855186333165,
      "grad_norm": 0.12166509653474573,
      "learning_rate": 9.086068781840521e-05,
      "loss": 1.0331,
      "step": 7537
    },
    {
      "epoch": 0.5433576010956535,
      "grad_norm": 0.1487389787662124,
      "learning_rate": 9.08374386269624e-05,
      "loss": 1.0734,
      "step": 7538
    },
    {
      "epoch": 0.5434296835579904,
      "grad_norm": 0.1223019884769976,
      "learning_rate": 9.081418993496114e-05,
      "loss": 1.0313,
      "step": 7539
    },
    {
      "epoch": 0.5435017660203273,
      "grad_norm": 0.258092288045683,
      "learning_rate": 9.079094174366867e-05,
      "loss": 1.0616,
      "step": 7540
    },
    {
      "epoch": 0.5435738484826642,
      "grad_norm": 0.14504575907012046,
      "learning_rate": 9.07676940543523e-05,
      "loss": 1.121,
      "step": 7541
    },
    {
      "epoch": 0.5436459309450011,
      "grad_norm": 0.1153271948330209,
      "learning_rate": 9.074444686827923e-05,
      "loss": 0.9407,
      "step": 7542
    },
    {
      "epoch": 0.5437180134073379,
      "grad_norm": 0.11861010590034711,
      "learning_rate": 9.072120018671657e-05,
      "loss": 1.0292,
      "step": 7543
    },
    {
      "epoch": 0.5437900958696749,
      "grad_norm": 0.11220985488894078,
      "learning_rate": 9.06979540109315e-05,
      "loss": 1.0914,
      "step": 7544
    },
    {
      "epoch": 0.5438621783320118,
      "grad_norm": 0.191914303524801,
      "learning_rate": 9.067470834219116e-05,
      "loss": 1.041,
      "step": 7545
    },
    {
      "epoch": 0.5439342607943487,
      "grad_norm": 0.11685779018681168,
      "learning_rate": 9.065146318176268e-05,
      "loss": 0.9819,
      "step": 7546
    },
    {
      "epoch": 0.5440063432566856,
      "grad_norm": 0.1402913981906928,
      "learning_rate": 9.062821853091305e-05,
      "loss": 1.0842,
      "step": 7547
    },
    {
      "epoch": 0.5440784257190225,
      "grad_norm": 0.13031101607252982,
      "learning_rate": 9.060497439090935e-05,
      "loss": 1.0554,
      "step": 7548
    },
    {
      "epoch": 0.5441505081813595,
      "grad_norm": 0.12804279260412607,
      "learning_rate": 9.058173076301861e-05,
      "loss": 1.1177,
      "step": 7549
    },
    {
      "epoch": 0.5442225906436964,
      "grad_norm": 0.11180397208513901,
      "learning_rate": 9.055848764850782e-05,
      "loss": 1.0029,
      "step": 7550
    },
    {
      "epoch": 0.5442946731060333,
      "grad_norm": 0.13214461692732393,
      "learning_rate": 9.05352450486439e-05,
      "loss": 1.0421,
      "step": 7551
    },
    {
      "epoch": 0.5443667555683702,
      "grad_norm": 0.13072931287125306,
      "learning_rate": 9.051200296469382e-05,
      "loss": 1.1206,
      "step": 7552
    },
    {
      "epoch": 0.5444388380307071,
      "grad_norm": 0.12843240003710715,
      "learning_rate": 9.048876139792441e-05,
      "loss": 1.0475,
      "step": 7553
    },
    {
      "epoch": 0.5445109204930441,
      "grad_norm": 0.11195977073814561,
      "learning_rate": 9.046552034960263e-05,
      "loss": 1.0728,
      "step": 7554
    },
    {
      "epoch": 0.544583002955381,
      "grad_norm": 0.12221291636557931,
      "learning_rate": 9.044227982099529e-05,
      "loss": 0.9576,
      "step": 7555
    },
    {
      "epoch": 0.5446550854177179,
      "grad_norm": 0.11466995588157092,
      "learning_rate": 9.041903981336922e-05,
      "loss": 1.0845,
      "step": 7556
    },
    {
      "epoch": 0.5447271678800548,
      "grad_norm": 0.11825386731428729,
      "learning_rate": 9.039580032799113e-05,
      "loss": 1.005,
      "step": 7557
    },
    {
      "epoch": 0.5447992503423917,
      "grad_norm": 0.1603331663259374,
      "learning_rate": 9.037256136612791e-05,
      "loss": 0.9943,
      "step": 7558
    },
    {
      "epoch": 0.5448713328047287,
      "grad_norm": 0.12826158657700099,
      "learning_rate": 9.03493229290462e-05,
      "loss": 1.0962,
      "step": 7559
    },
    {
      "epoch": 0.5449434152670656,
      "grad_norm": 0.09300528198475534,
      "learning_rate": 9.032608501801275e-05,
      "loss": 0.9452,
      "step": 7560
    },
    {
      "epoch": 0.5450154977294024,
      "grad_norm": 0.136757324745855,
      "learning_rate": 9.030284763429419e-05,
      "loss": 1.0434,
      "step": 7561
    },
    {
      "epoch": 0.5450875801917393,
      "grad_norm": 0.11032996349798294,
      "learning_rate": 9.027961077915717e-05,
      "loss": 0.9868,
      "step": 7562
    },
    {
      "epoch": 0.5451596626540762,
      "grad_norm": 0.23286157116055645,
      "learning_rate": 9.025637445386834e-05,
      "loss": 1.0119,
      "step": 7563
    },
    {
      "epoch": 0.5452317451164131,
      "grad_norm": 0.2602512200893916,
      "learning_rate": 9.02331386596943e-05,
      "loss": 1.0383,
      "step": 7564
    },
    {
      "epoch": 0.5453038275787501,
      "grad_norm": 0.18526893187227914,
      "learning_rate": 9.020990339790155e-05,
      "loss": 1.0099,
      "step": 7565
    },
    {
      "epoch": 0.545375910041087,
      "grad_norm": 0.1365646874419179,
      "learning_rate": 9.018666866975663e-05,
      "loss": 1.0632,
      "step": 7566
    },
    {
      "epoch": 0.5454479925034239,
      "grad_norm": 0.10681628041932423,
      "learning_rate": 9.016343447652608e-05,
      "loss": 1.0888,
      "step": 7567
    },
    {
      "epoch": 0.5455200749657608,
      "grad_norm": 0.12895918566215073,
      "learning_rate": 9.014020081947638e-05,
      "loss": 1.0225,
      "step": 7568
    },
    {
      "epoch": 0.5455921574280977,
      "grad_norm": 0.11938852799493424,
      "learning_rate": 9.011696769987392e-05,
      "loss": 1.0765,
      "step": 7569
    },
    {
      "epoch": 0.5456642398904347,
      "grad_norm": 0.11812763035034801,
      "learning_rate": 9.009373511898511e-05,
      "loss": 1.056,
      "step": 7570
    },
    {
      "epoch": 0.5457363223527716,
      "grad_norm": 0.1279147497520647,
      "learning_rate": 9.007050307807639e-05,
      "loss": 0.9906,
      "step": 7571
    },
    {
      "epoch": 0.5458084048151085,
      "grad_norm": 0.13295934972810267,
      "learning_rate": 9.00472715784141e-05,
      "loss": 1.0358,
      "step": 7572
    },
    {
      "epoch": 0.5458804872774454,
      "grad_norm": 0.1517572408273075,
      "learning_rate": 9.002404062126452e-05,
      "loss": 1.0458,
      "step": 7573
    },
    {
      "epoch": 0.5459525697397823,
      "grad_norm": 0.17772675922627687,
      "learning_rate": 9.0000810207894e-05,
      "loss": 1.0407,
      "step": 7574
    },
    {
      "epoch": 0.5460246522021193,
      "grad_norm": 0.17084678863456473,
      "learning_rate": 8.997758033956871e-05,
      "loss": 1.0892,
      "step": 7575
    },
    {
      "epoch": 0.5460967346644562,
      "grad_norm": 0.12037641393398328,
      "learning_rate": 8.995435101755503e-05,
      "loss": 1.0402,
      "step": 7576
    },
    {
      "epoch": 0.5461688171267931,
      "grad_norm": 0.16128793171570593,
      "learning_rate": 8.993112224311907e-05,
      "loss": 1.116,
      "step": 7577
    },
    {
      "epoch": 0.5462408995891299,
      "grad_norm": 0.13013786227150742,
      "learning_rate": 8.990789401752703e-05,
      "loss": 1.12,
      "step": 7578
    },
    {
      "epoch": 0.5463129820514668,
      "grad_norm": 1.5782832169976446,
      "learning_rate": 8.988466634204501e-05,
      "loss": 1.0648,
      "step": 7579
    },
    {
      "epoch": 0.5463850645138038,
      "grad_norm": 0.12124223084025638,
      "learning_rate": 8.986143921793923e-05,
      "loss": 1.0334,
      "step": 7580
    },
    {
      "epoch": 0.5464571469761407,
      "grad_norm": 0.11976033444063786,
      "learning_rate": 8.983821264647568e-05,
      "loss": 1.0946,
      "step": 7581
    },
    {
      "epoch": 0.5465292294384776,
      "grad_norm": 0.10614490472770717,
      "learning_rate": 8.981498662892048e-05,
      "loss": 1.1099,
      "step": 7582
    },
    {
      "epoch": 0.5466013119008145,
      "grad_norm": 0.20996718766884612,
      "learning_rate": 8.97917611665396e-05,
      "loss": 0.9832,
      "step": 7583
    },
    {
      "epoch": 0.5466733943631514,
      "grad_norm": 0.1692665393918795,
      "learning_rate": 8.976853626059905e-05,
      "loss": 1.0809,
      "step": 7584
    },
    {
      "epoch": 0.5467454768254884,
      "grad_norm": 0.14248690637811037,
      "learning_rate": 8.974531191236482e-05,
      "loss": 1.0372,
      "step": 7585
    },
    {
      "epoch": 0.5468175592878253,
      "grad_norm": 0.14378281512573704,
      "learning_rate": 8.972208812310286e-05,
      "loss": 1.0642,
      "step": 7586
    },
    {
      "epoch": 0.5468896417501622,
      "grad_norm": 0.114385500058911,
      "learning_rate": 8.969886489407901e-05,
      "loss": 1.0489,
      "step": 7587
    },
    {
      "epoch": 0.5469617242124991,
      "grad_norm": 0.2565789714087215,
      "learning_rate": 8.967564222655918e-05,
      "loss": 1.0766,
      "step": 7588
    },
    {
      "epoch": 0.547033806674836,
      "grad_norm": 0.12101340702054524,
      "learning_rate": 8.965242012180922e-05,
      "loss": 0.9853,
      "step": 7589
    },
    {
      "epoch": 0.547105889137173,
      "grad_norm": 0.12254259904171368,
      "learning_rate": 8.962919858109496e-05,
      "loss": 1.0249,
      "step": 7590
    },
    {
      "epoch": 0.5471779715995099,
      "grad_norm": 0.12430335386670165,
      "learning_rate": 8.960597760568215e-05,
      "loss": 1.0112,
      "step": 7591
    },
    {
      "epoch": 0.5472500540618468,
      "grad_norm": 0.13220894316428974,
      "learning_rate": 8.95827571968365e-05,
      "loss": 1.0889,
      "step": 7592
    },
    {
      "epoch": 0.5473221365241837,
      "grad_norm": 0.9439313765791381,
      "learning_rate": 8.955953735582382e-05,
      "loss": 1.0996,
      "step": 7593
    },
    {
      "epoch": 0.5473942189865206,
      "grad_norm": 0.1209297604997791,
      "learning_rate": 8.953631808390976e-05,
      "loss": 0.9764,
      "step": 7594
    },
    {
      "epoch": 0.5474663014488574,
      "grad_norm": 0.1312382241129483,
      "learning_rate": 8.951309938235998e-05,
      "loss": 1.1138,
      "step": 7595
    },
    {
      "epoch": 0.5475383839111944,
      "grad_norm": 0.1531363329895698,
      "learning_rate": 8.948988125244009e-05,
      "loss": 1.0987,
      "step": 7596
    },
    {
      "epoch": 0.5476104663735313,
      "grad_norm": 0.7811429588030396,
      "learning_rate": 8.946666369541567e-05,
      "loss": 1.1102,
      "step": 7597
    },
    {
      "epoch": 0.5476825488358682,
      "grad_norm": 0.14836662715119422,
      "learning_rate": 8.944344671255236e-05,
      "loss": 0.9409,
      "step": 7598
    },
    {
      "epoch": 0.5477546312982051,
      "grad_norm": 0.10703646308117404,
      "learning_rate": 8.942023030511562e-05,
      "loss": 1.0738,
      "step": 7599
    },
    {
      "epoch": 0.547826713760542,
      "grad_norm": 0.11738582347001282,
      "learning_rate": 8.939701447437097e-05,
      "loss": 0.9974,
      "step": 7600
    },
    {
      "epoch": 0.547898796222879,
      "grad_norm": 0.111939549183766,
      "learning_rate": 8.937379922158389e-05,
      "loss": 1.0752,
      "step": 7601
    },
    {
      "epoch": 0.5479708786852159,
      "grad_norm": 0.12514837519694202,
      "learning_rate": 8.935058454801984e-05,
      "loss": 1.0627,
      "step": 7602
    },
    {
      "epoch": 0.5480429611475528,
      "grad_norm": 0.11052379521269334,
      "learning_rate": 8.932737045494421e-05,
      "loss": 1.0429,
      "step": 7603
    },
    {
      "epoch": 0.5481150436098897,
      "grad_norm": 0.10852432333475716,
      "learning_rate": 8.930415694362235e-05,
      "loss": 1.04,
      "step": 7604
    },
    {
      "epoch": 0.5481871260722266,
      "grad_norm": 0.11568834975365816,
      "learning_rate": 8.928094401531964e-05,
      "loss": 1.0931,
      "step": 7605
    },
    {
      "epoch": 0.5482592085345636,
      "grad_norm": 0.11986713806074593,
      "learning_rate": 8.925773167130133e-05,
      "loss": 1.0664,
      "step": 7606
    },
    {
      "epoch": 0.5483312909969005,
      "grad_norm": 0.12134716176878292,
      "learning_rate": 8.923451991283279e-05,
      "loss": 1.0091,
      "step": 7607
    },
    {
      "epoch": 0.5484033734592374,
      "grad_norm": 0.13463545072645589,
      "learning_rate": 8.921130874117921e-05,
      "loss": 1.0668,
      "step": 7608
    },
    {
      "epoch": 0.5484754559215743,
      "grad_norm": 0.11141644379947879,
      "learning_rate": 8.918809815760585e-05,
      "loss": 1.1239,
      "step": 7609
    },
    {
      "epoch": 0.5485475383839112,
      "grad_norm": 0.1334250144233733,
      "learning_rate": 8.916488816337782e-05,
      "loss": 1.0291,
      "step": 7610
    },
    {
      "epoch": 0.5486196208462482,
      "grad_norm": 0.16783132041314217,
      "learning_rate": 8.914167875976034e-05,
      "loss": 1.1704,
      "step": 7611
    },
    {
      "epoch": 0.548691703308585,
      "grad_norm": 0.1664561430889062,
      "learning_rate": 8.911846994801851e-05,
      "loss": 1.0541,
      "step": 7612
    },
    {
      "epoch": 0.5487637857709219,
      "grad_norm": 0.13049833709345043,
      "learning_rate": 8.909526172941745e-05,
      "loss": 1.0974,
      "step": 7613
    },
    {
      "epoch": 0.5488358682332588,
      "grad_norm": 0.11217159237512118,
      "learning_rate": 8.907205410522212e-05,
      "loss": 1.0694,
      "step": 7614
    },
    {
      "epoch": 0.5489079506955957,
      "grad_norm": 0.11427045899903511,
      "learning_rate": 8.904884707669764e-05,
      "loss": 1.0325,
      "step": 7615
    },
    {
      "epoch": 0.5489800331579326,
      "grad_norm": 0.10855199066786425,
      "learning_rate": 8.902564064510897e-05,
      "loss": 0.9952,
      "step": 7616
    },
    {
      "epoch": 0.5490521156202696,
      "grad_norm": 0.32014286221774385,
      "learning_rate": 8.900243481172107e-05,
      "loss": 1.0192,
      "step": 7617
    },
    {
      "epoch": 0.5491241980826065,
      "grad_norm": 0.12226295527381272,
      "learning_rate": 8.897922957779885e-05,
      "loss": 1.1067,
      "step": 7618
    },
    {
      "epoch": 0.5491962805449434,
      "grad_norm": 0.11400838932681341,
      "learning_rate": 8.895602494460719e-05,
      "loss": 1.0183,
      "step": 7619
    },
    {
      "epoch": 0.5492683630072803,
      "grad_norm": 0.1163551017067318,
      "learning_rate": 8.893282091341099e-05,
      "loss": 1.003,
      "step": 7620
    },
    {
      "epoch": 0.5493404454696172,
      "grad_norm": 0.11563095680351708,
      "learning_rate": 8.890961748547508e-05,
      "loss": 1.0373,
      "step": 7621
    },
    {
      "epoch": 0.5494125279319542,
      "grad_norm": 0.1527177750926483,
      "learning_rate": 8.888641466206421e-05,
      "loss": 0.9549,
      "step": 7622
    },
    {
      "epoch": 0.5494846103942911,
      "grad_norm": 0.11637585714096131,
      "learning_rate": 8.886321244444316e-05,
      "loss": 1.0312,
      "step": 7623
    },
    {
      "epoch": 0.549556692856628,
      "grad_norm": 0.11773708984520787,
      "learning_rate": 8.884001083387667e-05,
      "loss": 1.0058,
      "step": 7624
    },
    {
      "epoch": 0.5496287753189649,
      "grad_norm": 0.11494261140443064,
      "learning_rate": 8.881680983162946e-05,
      "loss": 1.1073,
      "step": 7625
    },
    {
      "epoch": 0.5497008577813018,
      "grad_norm": 0.4330968577708169,
      "learning_rate": 8.879360943896613e-05,
      "loss": 1.0982,
      "step": 7626
    },
    {
      "epoch": 0.5497729402436388,
      "grad_norm": 0.10777965513218261,
      "learning_rate": 8.877040965715137e-05,
      "loss": 1.0991,
      "step": 7627
    },
    {
      "epoch": 0.5498450227059757,
      "grad_norm": 0.19585909095627033,
      "learning_rate": 8.874721048744968e-05,
      "loss": 1.0687,
      "step": 7628
    },
    {
      "epoch": 0.5499171051683126,
      "grad_norm": 1.2632538395715673,
      "learning_rate": 8.872401193112576e-05,
      "loss": 0.993,
      "step": 7629
    },
    {
      "epoch": 0.5499891876306494,
      "grad_norm": 0.12234527750585174,
      "learning_rate": 8.870081398944405e-05,
      "loss": 1.0446,
      "step": 7630
    },
    {
      "epoch": 0.5500612700929863,
      "grad_norm": 0.127612451256894,
      "learning_rate": 8.867761666366907e-05,
      "loss": 1.0608,
      "step": 7631
    },
    {
      "epoch": 0.5501333525553233,
      "grad_norm": 0.14405002870753386,
      "learning_rate": 8.86544199550652e-05,
      "loss": 1.0585,
      "step": 7632
    },
    {
      "epoch": 0.5502054350176602,
      "grad_norm": 0.11421079899751525,
      "learning_rate": 8.863122386489704e-05,
      "loss": 1.0267,
      "step": 7633
    },
    {
      "epoch": 0.5502775174799971,
      "grad_norm": 0.12486159497586577,
      "learning_rate": 8.860802839442885e-05,
      "loss": 1.1332,
      "step": 7634
    },
    {
      "epoch": 0.550349599942334,
      "grad_norm": 0.11952750964653623,
      "learning_rate": 8.858483354492504e-05,
      "loss": 1.0259,
      "step": 7635
    },
    {
      "epoch": 0.5504216824046709,
      "grad_norm": 0.12887792883034482,
      "learning_rate": 8.856163931764987e-05,
      "loss": 1.0793,
      "step": 7636
    },
    {
      "epoch": 0.5504937648670079,
      "grad_norm": 0.16674977532894886,
      "learning_rate": 8.853844571386776e-05,
      "loss": 1.1215,
      "step": 7637
    },
    {
      "epoch": 0.5505658473293448,
      "grad_norm": 0.297943066919298,
      "learning_rate": 8.851525273484286e-05,
      "loss": 1.0914,
      "step": 7638
    },
    {
      "epoch": 0.5506379297916817,
      "grad_norm": 0.12064718001153228,
      "learning_rate": 8.849206038183945e-05,
      "loss": 1.0336,
      "step": 7639
    },
    {
      "epoch": 0.5507100122540186,
      "grad_norm": 0.12250988934758278,
      "learning_rate": 8.846886865612169e-05,
      "loss": 1.0272,
      "step": 7640
    },
    {
      "epoch": 0.5507820947163555,
      "grad_norm": 0.1165471012283468,
      "learning_rate": 8.844567755895374e-05,
      "loss": 0.9236,
      "step": 7641
    },
    {
      "epoch": 0.5508541771786925,
      "grad_norm": 0.26524824962102556,
      "learning_rate": 8.842248709159972e-05,
      "loss": 1.044,
      "step": 7642
    },
    {
      "epoch": 0.5509262596410294,
      "grad_norm": 0.16198514323091157,
      "learning_rate": 8.839929725532378e-05,
      "loss": 1.0158,
      "step": 7643
    },
    {
      "epoch": 0.5509983421033663,
      "grad_norm": 0.14665394033738074,
      "learning_rate": 8.837610805138989e-05,
      "loss": 1.0409,
      "step": 7644
    },
    {
      "epoch": 0.5510704245657032,
      "grad_norm": 0.12460901801244044,
      "learning_rate": 8.835291948106207e-05,
      "loss": 1.095,
      "step": 7645
    },
    {
      "epoch": 0.5511425070280401,
      "grad_norm": 0.17808832424203805,
      "learning_rate": 8.832973154560438e-05,
      "loss": 1.0173,
      "step": 7646
    },
    {
      "epoch": 0.5512145894903769,
      "grad_norm": 0.12824827939812108,
      "learning_rate": 8.830654424628074e-05,
      "loss": 1.1082,
      "step": 7647
    },
    {
      "epoch": 0.5512866719527139,
      "grad_norm": 0.12193854751478203,
      "learning_rate": 8.828335758435502e-05,
      "loss": 1.0616,
      "step": 7648
    },
    {
      "epoch": 0.5513587544150508,
      "grad_norm": 0.13117789880101302,
      "learning_rate": 8.826017156109116e-05,
      "loss": 1.1297,
      "step": 7649
    },
    {
      "epoch": 0.5514308368773877,
      "grad_norm": 0.11438065986970589,
      "learning_rate": 8.823698617775292e-05,
      "loss": 1.1087,
      "step": 7650
    },
    {
      "epoch": 0.5515029193397246,
      "grad_norm": 0.7763540301416075,
      "learning_rate": 8.821380143560422e-05,
      "loss": 1.0627,
      "step": 7651
    },
    {
      "epoch": 0.5515750018020615,
      "grad_norm": 0.13562618333013954,
      "learning_rate": 8.819061733590876e-05,
      "loss": 1.0487,
      "step": 7652
    },
    {
      "epoch": 0.5516470842643985,
      "grad_norm": 0.13987788825853448,
      "learning_rate": 8.816743387993034e-05,
      "loss": 1.0786,
      "step": 7653
    },
    {
      "epoch": 0.5517191667267354,
      "grad_norm": 0.13117501841869922,
      "learning_rate": 8.814425106893256e-05,
      "loss": 1.0342,
      "step": 7654
    },
    {
      "epoch": 0.5517912491890723,
      "grad_norm": 0.10602425543702501,
      "learning_rate": 8.812106890417923e-05,
      "loss": 0.9923,
      "step": 7655
    },
    {
      "epoch": 0.5518633316514092,
      "grad_norm": 0.18362943392850706,
      "learning_rate": 8.809788738693388e-05,
      "loss": 1.0458,
      "step": 7656
    },
    {
      "epoch": 0.5519354141137461,
      "grad_norm": 0.1214082593920457,
      "learning_rate": 8.807470651846018e-05,
      "loss": 1.1003,
      "step": 7657
    },
    {
      "epoch": 0.5520074965760831,
      "grad_norm": 0.1550585656133664,
      "learning_rate": 8.80515263000216e-05,
      "loss": 1.0496,
      "step": 7658
    },
    {
      "epoch": 0.55207957903842,
      "grad_norm": 0.1270265057088052,
      "learning_rate": 8.802834673288178e-05,
      "loss": 1.0328,
      "step": 7659
    },
    {
      "epoch": 0.5521516615007569,
      "grad_norm": 0.11886778330428807,
      "learning_rate": 8.800516781830417e-05,
      "loss": 1.0297,
      "step": 7660
    },
    {
      "epoch": 0.5522237439630938,
      "grad_norm": 0.11363259847027513,
      "learning_rate": 8.798198955755223e-05,
      "loss": 1.0988,
      "step": 7661
    },
    {
      "epoch": 0.5522958264254307,
      "grad_norm": 0.10712372524031702,
      "learning_rate": 8.795881195188934e-05,
      "loss": 1.0337,
      "step": 7662
    },
    {
      "epoch": 0.5523679088877677,
      "grad_norm": 0.12242633231470568,
      "learning_rate": 8.793563500257893e-05,
      "loss": 1.0838,
      "step": 7663
    },
    {
      "epoch": 0.5524399913501045,
      "grad_norm": 0.12446818423060106,
      "learning_rate": 8.791245871088433e-05,
      "loss": 1.0738,
      "step": 7664
    },
    {
      "epoch": 0.5525120738124414,
      "grad_norm": 0.11164351401931764,
      "learning_rate": 8.78892830780689e-05,
      "loss": 1.0623,
      "step": 7665
    },
    {
      "epoch": 0.5525841562747783,
      "grad_norm": 0.11979649039615037,
      "learning_rate": 8.786610810539588e-05,
      "loss": 1.0503,
      "step": 7666
    },
    {
      "epoch": 0.5526562387371152,
      "grad_norm": 0.1410343712784529,
      "learning_rate": 8.78429337941285e-05,
      "loss": 1.0575,
      "step": 7667
    },
    {
      "epoch": 0.5527283211994521,
      "grad_norm": 0.11036734455551762,
      "learning_rate": 8.781976014553001e-05,
      "loss": 1.0153,
      "step": 7668
    },
    {
      "epoch": 0.5528004036617891,
      "grad_norm": 0.12292785304022652,
      "learning_rate": 8.779658716086356e-05,
      "loss": 1.0209,
      "step": 7669
    },
    {
      "epoch": 0.552872486124126,
      "grad_norm": 0.11492064417558095,
      "learning_rate": 8.777341484139229e-05,
      "loss": 1.1072,
      "step": 7670
    },
    {
      "epoch": 0.5529445685864629,
      "grad_norm": 0.13527878365659338,
      "learning_rate": 8.775024318837928e-05,
      "loss": 1.0205,
      "step": 7671
    },
    {
      "epoch": 0.5530166510487998,
      "grad_norm": 0.1283160998161735,
      "learning_rate": 8.772707220308758e-05,
      "loss": 1.056,
      "step": 7672
    },
    {
      "epoch": 0.5530887335111367,
      "grad_norm": 0.12091679668233364,
      "learning_rate": 8.770390188678029e-05,
      "loss": 0.9995,
      "step": 7673
    },
    {
      "epoch": 0.5531608159734737,
      "grad_norm": 0.14417989100103354,
      "learning_rate": 8.768073224072033e-05,
      "loss": 1.0748,
      "step": 7674
    },
    {
      "epoch": 0.5532328984358106,
      "grad_norm": 0.13377487497863802,
      "learning_rate": 8.765756326617068e-05,
      "loss": 1.0478,
      "step": 7675
    },
    {
      "epoch": 0.5533049808981475,
      "grad_norm": 0.2698847805536831,
      "learning_rate": 8.763439496439422e-05,
      "loss": 1.0095,
      "step": 7676
    },
    {
      "epoch": 0.5533770633604844,
      "grad_norm": 0.11318782341881194,
      "learning_rate": 8.76112273366539e-05,
      "loss": 1.01,
      "step": 7677
    },
    {
      "epoch": 0.5534491458228213,
      "grad_norm": 0.11610341018976793,
      "learning_rate": 8.758806038421251e-05,
      "loss": 1.1143,
      "step": 7678
    },
    {
      "epoch": 0.5535212282851583,
      "grad_norm": 0.11203474200761547,
      "learning_rate": 8.756489410833287e-05,
      "loss": 1.0262,
      "step": 7679
    },
    {
      "epoch": 0.5535933107474952,
      "grad_norm": 0.12249567510293409,
      "learning_rate": 8.754172851027772e-05,
      "loss": 1.0613,
      "step": 7680
    },
    {
      "epoch": 0.5536653932098321,
      "grad_norm": 0.10522798021805993,
      "learning_rate": 8.751856359130988e-05,
      "loss": 1.0863,
      "step": 7681
    },
    {
      "epoch": 0.5537374756721689,
      "grad_norm": 0.09852702350282018,
      "learning_rate": 8.749539935269195e-05,
      "loss": 1.0324,
      "step": 7682
    },
    {
      "epoch": 0.5538095581345058,
      "grad_norm": 0.11512480292382549,
      "learning_rate": 8.747223579568662e-05,
      "loss": 0.9723,
      "step": 7683
    },
    {
      "epoch": 0.5538816405968428,
      "grad_norm": 0.18098902178318402,
      "learning_rate": 8.744907292155655e-05,
      "loss": 1.049,
      "step": 7684
    },
    {
      "epoch": 0.5539537230591797,
      "grad_norm": 0.10856922753674021,
      "learning_rate": 8.742591073156424e-05,
      "loss": 1.0337,
      "step": 7685
    },
    {
      "epoch": 0.5540258055215166,
      "grad_norm": 0.12305915581206808,
      "learning_rate": 8.740274922697231e-05,
      "loss": 1.013,
      "step": 7686
    },
    {
      "epoch": 0.5540978879838535,
      "grad_norm": 0.3519798002566445,
      "learning_rate": 8.737958840904324e-05,
      "loss": 1.0674,
      "step": 7687
    },
    {
      "epoch": 0.5541699704461904,
      "grad_norm": 0.1292230444213562,
      "learning_rate": 8.735642827903953e-05,
      "loss": 1.0258,
      "step": 7688
    },
    {
      "epoch": 0.5542420529085273,
      "grad_norm": 0.13508007231274158,
      "learning_rate": 8.733326883822356e-05,
      "loss": 1.1883,
      "step": 7689
    },
    {
      "epoch": 0.5543141353708643,
      "grad_norm": 0.12401733122792832,
      "learning_rate": 8.731011008785776e-05,
      "loss": 1.0737,
      "step": 7690
    },
    {
      "epoch": 0.5543862178332012,
      "grad_norm": 0.17853024984316113,
      "learning_rate": 8.728695202920451e-05,
      "loss": 1.0352,
      "step": 7691
    },
    {
      "epoch": 0.5544583002955381,
      "grad_norm": 0.12775988812911984,
      "learning_rate": 8.726379466352613e-05,
      "loss": 1.0788,
      "step": 7692
    },
    {
      "epoch": 0.554530382757875,
      "grad_norm": 0.11246627809654688,
      "learning_rate": 8.724063799208483e-05,
      "loss": 0.9883,
      "step": 7693
    },
    {
      "epoch": 0.554602465220212,
      "grad_norm": 0.1443349494533032,
      "learning_rate": 8.721748201614292e-05,
      "loss": 1.0037,
      "step": 7694
    },
    {
      "epoch": 0.5546745476825489,
      "grad_norm": 0.11575045171363141,
      "learning_rate": 8.719432673696259e-05,
      "loss": 1.0362,
      "step": 7695
    },
    {
      "epoch": 0.5547466301448858,
      "grad_norm": 0.14359452410466728,
      "learning_rate": 8.717117215580606e-05,
      "loss": 1.1227,
      "step": 7696
    },
    {
      "epoch": 0.5548187126072227,
      "grad_norm": 0.20487219413188149,
      "learning_rate": 8.714801827393538e-05,
      "loss": 1.063,
      "step": 7697
    },
    {
      "epoch": 0.5548907950695596,
      "grad_norm": 0.12249382005099477,
      "learning_rate": 8.712486509261267e-05,
      "loss": 1.0811,
      "step": 7698
    },
    {
      "epoch": 0.5549628775318964,
      "grad_norm": 0.10962784101221132,
      "learning_rate": 8.710171261310002e-05,
      "loss": 1.0574,
      "step": 7699
    },
    {
      "epoch": 0.5550349599942334,
      "grad_norm": 0.19257251343451667,
      "learning_rate": 8.707856083665944e-05,
      "loss": 1.0456,
      "step": 7700
    },
    {
      "epoch": 0.5551070424565703,
      "grad_norm": 0.16643112873378027,
      "learning_rate": 8.705540976455287e-05,
      "loss": 0.9858,
      "step": 7701
    },
    {
      "epoch": 0.5551791249189072,
      "grad_norm": 0.1208225626254918,
      "learning_rate": 8.703225939804225e-05,
      "loss": 1.0487,
      "step": 7702
    },
    {
      "epoch": 0.5552512073812441,
      "grad_norm": 0.38053162238514615,
      "learning_rate": 8.700910973838952e-05,
      "loss": 1.0609,
      "step": 7703
    },
    {
      "epoch": 0.555323289843581,
      "grad_norm": 0.13972784420931988,
      "learning_rate": 8.698596078685656e-05,
      "loss": 1.1233,
      "step": 7704
    },
    {
      "epoch": 0.555395372305918,
      "grad_norm": 0.11503292266781165,
      "learning_rate": 8.696281254470516e-05,
      "loss": 1.0651,
      "step": 7705
    },
    {
      "epoch": 0.5554674547682549,
      "grad_norm": 0.15035705107879677,
      "learning_rate": 8.693966501319711e-05,
      "loss": 1.0903,
      "step": 7706
    },
    {
      "epoch": 0.5555395372305918,
      "grad_norm": 0.11500876558620146,
      "learning_rate": 8.691651819359409e-05,
      "loss": 1.0817,
      "step": 7707
    },
    {
      "epoch": 0.5556116196929287,
      "grad_norm": 0.4042053203902156,
      "learning_rate": 8.689337208715796e-05,
      "loss": 1.1312,
      "step": 7708
    },
    {
      "epoch": 0.5556837021552656,
      "grad_norm": 0.14217109982360998,
      "learning_rate": 8.687022669515027e-05,
      "loss": 0.9921,
      "step": 7709
    },
    {
      "epoch": 0.5557557846176026,
      "grad_norm": 0.11108106818322833,
      "learning_rate": 8.684708201883272e-05,
      "loss": 1.0684,
      "step": 7710
    },
    {
      "epoch": 0.5558278670799395,
      "grad_norm": 0.120036136944038,
      "learning_rate": 8.682393805946679e-05,
      "loss": 1.0146,
      "step": 7711
    },
    {
      "epoch": 0.5558999495422764,
      "grad_norm": 1.703677171476257,
      "learning_rate": 8.680079481831419e-05,
      "loss": 1.1357,
      "step": 7712
    },
    {
      "epoch": 0.5559720320046133,
      "grad_norm": 0.10974886332614349,
      "learning_rate": 8.677765229663633e-05,
      "loss": 1.1082,
      "step": 7713
    },
    {
      "epoch": 0.5560441144669502,
      "grad_norm": 0.762472347596575,
      "learning_rate": 8.675451049569472e-05,
      "loss": 1.0848,
      "step": 7714
    },
    {
      "epoch": 0.5561161969292872,
      "grad_norm": 2.8099272298574385,
      "learning_rate": 8.673136941675079e-05,
      "loss": 1.0605,
      "step": 7715
    },
    {
      "epoch": 0.556188279391624,
      "grad_norm": 0.1446670121228274,
      "learning_rate": 8.670822906106589e-05,
      "loss": 1.0438,
      "step": 7716
    },
    {
      "epoch": 0.5562603618539609,
      "grad_norm": 0.12227496674716551,
      "learning_rate": 8.668508942990145e-05,
      "loss": 1.0664,
      "step": 7717
    },
    {
      "epoch": 0.5563324443162978,
      "grad_norm": 0.11206959626013288,
      "learning_rate": 8.666195052451877e-05,
      "loss": 1.0392,
      "step": 7718
    },
    {
      "epoch": 0.5564045267786347,
      "grad_norm": 0.11646548494969633,
      "learning_rate": 8.663881234617909e-05,
      "loss": 1.0359,
      "step": 7719
    },
    {
      "epoch": 0.5564766092409716,
      "grad_norm": 0.10978992588827881,
      "learning_rate": 8.661567489614363e-05,
      "loss": 1.1458,
      "step": 7720
    },
    {
      "epoch": 0.5565486917033086,
      "grad_norm": 0.1334693368720806,
      "learning_rate": 8.659253817567369e-05,
      "loss": 1.1126,
      "step": 7721
    },
    {
      "epoch": 0.5566207741656455,
      "grad_norm": 0.10502500521901251,
      "learning_rate": 8.656940218603036e-05,
      "loss": 0.9472,
      "step": 7722
    },
    {
      "epoch": 0.5566928566279824,
      "grad_norm": 0.2336755813063564,
      "learning_rate": 8.654626692847475e-05,
      "loss": 1.1187,
      "step": 7723
    },
    {
      "epoch": 0.5567649390903193,
      "grad_norm": 0.12738424722724975,
      "learning_rate": 8.652313240426797e-05,
      "loss": 1.0023,
      "step": 7724
    },
    {
      "epoch": 0.5568370215526562,
      "grad_norm": 0.11761673757678963,
      "learning_rate": 8.649999861467099e-05,
      "loss": 1.0301,
      "step": 7725
    },
    {
      "epoch": 0.5569091040149932,
      "grad_norm": 0.1380738999413219,
      "learning_rate": 8.64768655609449e-05,
      "loss": 1.0469,
      "step": 7726
    },
    {
      "epoch": 0.5569811864773301,
      "grad_norm": 0.1264735872956914,
      "learning_rate": 8.645373324435062e-05,
      "loss": 0.9757,
      "step": 7727
    },
    {
      "epoch": 0.557053268939667,
      "grad_norm": 0.15777392185173167,
      "learning_rate": 8.643060166614908e-05,
      "loss": 1.0595,
      "step": 7728
    },
    {
      "epoch": 0.5571253514020039,
      "grad_norm": 0.3296366039259717,
      "learning_rate": 8.640747082760108e-05,
      "loss": 0.9756,
      "step": 7729
    },
    {
      "epoch": 0.5571974338643408,
      "grad_norm": 0.14072740867397446,
      "learning_rate": 8.638434072996759e-05,
      "loss": 1.0584,
      "step": 7730
    },
    {
      "epoch": 0.5572695163266778,
      "grad_norm": 0.24904540716738552,
      "learning_rate": 8.636121137450931e-05,
      "loss": 1.0574,
      "step": 7731
    },
    {
      "epoch": 0.5573415987890147,
      "grad_norm": 0.15340420430173216,
      "learning_rate": 8.633808276248705e-05,
      "loss": 1.0677,
      "step": 7732
    },
    {
      "epoch": 0.5574136812513516,
      "grad_norm": 0.16421791423087934,
      "learning_rate": 8.631495489516144e-05,
      "loss": 1.0085,
      "step": 7733
    },
    {
      "epoch": 0.5574857637136884,
      "grad_norm": 0.24398872957627343,
      "learning_rate": 8.629182777379328e-05,
      "loss": 1.071,
      "step": 7734
    },
    {
      "epoch": 0.5575578461760253,
      "grad_norm": 0.13430917797169165,
      "learning_rate": 8.626870139964313e-05,
      "loss": 1.0888,
      "step": 7735
    },
    {
      "epoch": 0.5576299286383622,
      "grad_norm": 0.13420181770323286,
      "learning_rate": 8.62455757739716e-05,
      "loss": 1.0277,
      "step": 7736
    },
    {
      "epoch": 0.5577020111006992,
      "grad_norm": 0.16092737910859553,
      "learning_rate": 8.622245089803925e-05,
      "loss": 1.0275,
      "step": 7737
    },
    {
      "epoch": 0.5577740935630361,
      "grad_norm": 0.1675353808786824,
      "learning_rate": 8.619932677310653e-05,
      "loss": 0.9899,
      "step": 7738
    },
    {
      "epoch": 0.557846176025373,
      "grad_norm": 0.16013893250206843,
      "learning_rate": 8.617620340043401e-05,
      "loss": 1.0408,
      "step": 7739
    },
    {
      "epoch": 0.5579182584877099,
      "grad_norm": 0.17204931667844578,
      "learning_rate": 8.61530807812821e-05,
      "loss": 1.0248,
      "step": 7740
    },
    {
      "epoch": 0.5579903409500468,
      "grad_norm": 0.10363382228992829,
      "learning_rate": 8.612995891691113e-05,
      "loss": 1.0022,
      "step": 7741
    },
    {
      "epoch": 0.5580624234123838,
      "grad_norm": 0.11695387164187429,
      "learning_rate": 8.610683780858147e-05,
      "loss": 0.9819,
      "step": 7742
    },
    {
      "epoch": 0.5581345058747207,
      "grad_norm": 0.18255037797180548,
      "learning_rate": 8.608371745755345e-05,
      "loss": 1.0132,
      "step": 7743
    },
    {
      "epoch": 0.5582065883370576,
      "grad_norm": 0.13575782827485602,
      "learning_rate": 8.606059786508735e-05,
      "loss": 1.0629,
      "step": 7744
    },
    {
      "epoch": 0.5582786707993945,
      "grad_norm": 0.10735953914515005,
      "learning_rate": 8.603747903244335e-05,
      "loss": 1.0107,
      "step": 7745
    },
    {
      "epoch": 0.5583507532617314,
      "grad_norm": 0.1240463999338339,
      "learning_rate": 8.601436096088166e-05,
      "loss": 1.0318,
      "step": 7746
    },
    {
      "epoch": 0.5584228357240684,
      "grad_norm": 0.1307212621543906,
      "learning_rate": 8.59912436516624e-05,
      "loss": 1.069,
      "step": 7747
    },
    {
      "epoch": 0.5584949181864053,
      "grad_norm": 0.1622612922728412,
      "learning_rate": 8.596812710604574e-05,
      "loss": 1.0603,
      "step": 7748
    },
    {
      "epoch": 0.5585670006487422,
      "grad_norm": 0.14885374050816186,
      "learning_rate": 8.594501132529165e-05,
      "loss": 1.04,
      "step": 7749
    },
    {
      "epoch": 0.5586390831110791,
      "grad_norm": 0.11657739326974019,
      "learning_rate": 8.592189631066018e-05,
      "loss": 1.0552,
      "step": 7750
    },
    {
      "epoch": 0.5587111655734159,
      "grad_norm": 0.11291831893796023,
      "learning_rate": 8.589878206341129e-05,
      "loss": 1.0151,
      "step": 7751
    },
    {
      "epoch": 0.5587832480357529,
      "grad_norm": 0.10989885501555641,
      "learning_rate": 8.587566858480496e-05,
      "loss": 1.0582,
      "step": 7752
    },
    {
      "epoch": 0.5588553304980898,
      "grad_norm": 0.29686910348140366,
      "learning_rate": 8.585255587610105e-05,
      "loss": 1.0274,
      "step": 7753
    },
    {
      "epoch": 0.5589274129604267,
      "grad_norm": 0.12691040111062504,
      "learning_rate": 8.582944393855941e-05,
      "loss": 1.0076,
      "step": 7754
    },
    {
      "epoch": 0.5589994954227636,
      "grad_norm": 0.22484842292824855,
      "learning_rate": 8.580633277343982e-05,
      "loss": 0.9896,
      "step": 7755
    },
    {
      "epoch": 0.5590715778851005,
      "grad_norm": 0.11680256640982796,
      "learning_rate": 8.578322238200211e-05,
      "loss": 0.9709,
      "step": 7756
    },
    {
      "epoch": 0.5591436603474375,
      "grad_norm": 0.1047262876904005,
      "learning_rate": 8.576011276550596e-05,
      "loss": 1.0887,
      "step": 7757
    },
    {
      "epoch": 0.5592157428097744,
      "grad_norm": 0.18011558249900964,
      "learning_rate": 8.573700392521106e-05,
      "loss": 1.0413,
      "step": 7758
    },
    {
      "epoch": 0.5592878252721113,
      "grad_norm": 0.12193297819063298,
      "learning_rate": 8.571389586237706e-05,
      "loss": 1.0512,
      "step": 7759
    },
    {
      "epoch": 0.5593599077344482,
      "grad_norm": 0.1272822267664974,
      "learning_rate": 8.56907885782635e-05,
      "loss": 0.9973,
      "step": 7760
    },
    {
      "epoch": 0.5594319901967851,
      "grad_norm": 0.11975585175991349,
      "learning_rate": 8.566768207413e-05,
      "loss": 1.0778,
      "step": 7761
    },
    {
      "epoch": 0.559504072659122,
      "grad_norm": 0.11978263490098832,
      "learning_rate": 8.564457635123603e-05,
      "loss": 1.0727,
      "step": 7762
    },
    {
      "epoch": 0.559576155121459,
      "grad_norm": 0.19317151679538636,
      "learning_rate": 8.562147141084112e-05,
      "loss": 1.0404,
      "step": 7763
    },
    {
      "epoch": 0.5596482375837959,
      "grad_norm": 0.22042965825071908,
      "learning_rate": 8.55983672542046e-05,
      "loss": 1.0608,
      "step": 7764
    },
    {
      "epoch": 0.5597203200461328,
      "grad_norm": 0.14352239064011826,
      "learning_rate": 8.557526388258593e-05,
      "loss": 1.0868,
      "step": 7765
    },
    {
      "epoch": 0.5597924025084697,
      "grad_norm": 0.11713530888267215,
      "learning_rate": 8.555216129724442e-05,
      "loss": 1.0043,
      "step": 7766
    },
    {
      "epoch": 0.5598644849708067,
      "grad_norm": 0.11998111197576589,
      "learning_rate": 8.55290594994394e-05,
      "loss": 1.0526,
      "step": 7767
    },
    {
      "epoch": 0.5599365674331435,
      "grad_norm": 0.14352263387119135,
      "learning_rate": 8.550595849043008e-05,
      "loss": 0.9503,
      "step": 7768
    },
    {
      "epoch": 0.5600086498954804,
      "grad_norm": 0.12682863083858115,
      "learning_rate": 8.548285827147568e-05,
      "loss": 1.0513,
      "step": 7769
    },
    {
      "epoch": 0.5600807323578173,
      "grad_norm": 0.17322863190805943,
      "learning_rate": 8.545975884383537e-05,
      "loss": 1.0599,
      "step": 7770
    },
    {
      "epoch": 0.5601528148201542,
      "grad_norm": 0.1436696074782204,
      "learning_rate": 8.543666020876832e-05,
      "loss": 1.0014,
      "step": 7771
    },
    {
      "epoch": 0.5602248972824911,
      "grad_norm": 0.14706982004714345,
      "learning_rate": 8.541356236753357e-05,
      "loss": 1.0581,
      "step": 7772
    },
    {
      "epoch": 0.5602969797448281,
      "grad_norm": 0.11542996876285398,
      "learning_rate": 8.539046532139011e-05,
      "loss": 1.0537,
      "step": 7773
    },
    {
      "epoch": 0.560369062207165,
      "grad_norm": 0.12150323028253635,
      "learning_rate": 8.536736907159704e-05,
      "loss": 1.0313,
      "step": 7774
    },
    {
      "epoch": 0.5604411446695019,
      "grad_norm": 3.5393112808051894,
      "learning_rate": 8.534427361941327e-05,
      "loss": 1.1909,
      "step": 7775
    },
    {
      "epoch": 0.5605132271318388,
      "grad_norm": 0.37553358476878324,
      "learning_rate": 8.532117896609768e-05,
      "loss": 1.0604,
      "step": 7776
    },
    {
      "epoch": 0.5605853095941757,
      "grad_norm": 0.11516118535043943,
      "learning_rate": 8.529808511290915e-05,
      "loss": 1.0049,
      "step": 7777
    },
    {
      "epoch": 0.5606573920565127,
      "grad_norm": 0.4506778375349481,
      "learning_rate": 8.527499206110652e-05,
      "loss": 1.0998,
      "step": 7778
    },
    {
      "epoch": 0.5607294745188496,
      "grad_norm": 0.11364784976998746,
      "learning_rate": 8.525189981194857e-05,
      "loss": 1.0614,
      "step": 7779
    },
    {
      "epoch": 0.5608015569811865,
      "grad_norm": 0.12050248411247952,
      "learning_rate": 8.522880836669401e-05,
      "loss": 1.1338,
      "step": 7780
    },
    {
      "epoch": 0.5608736394435234,
      "grad_norm": 0.12173163656483836,
      "learning_rate": 8.520571772660155e-05,
      "loss": 1.0389,
      "step": 7781
    },
    {
      "epoch": 0.5609457219058603,
      "grad_norm": 0.16387542140769495,
      "learning_rate": 8.518262789292977e-05,
      "loss": 1.0708,
      "step": 7782
    },
    {
      "epoch": 0.5610178043681973,
      "grad_norm": 0.11608649398274326,
      "learning_rate": 8.515953886693739e-05,
      "loss": 1.0148,
      "step": 7783
    },
    {
      "epoch": 0.5610898868305342,
      "grad_norm": 0.11523521764699991,
      "learning_rate": 8.513645064988291e-05,
      "loss": 1.0573,
      "step": 7784
    },
    {
      "epoch": 0.5611619692928711,
      "grad_norm": 0.1288318106206309,
      "learning_rate": 8.511336324302483e-05,
      "loss": 1.045,
      "step": 7785
    },
    {
      "epoch": 0.5612340517552079,
      "grad_norm": 0.16177335812743557,
      "learning_rate": 8.509027664762159e-05,
      "loss": 1.0264,
      "step": 7786
    },
    {
      "epoch": 0.5613061342175448,
      "grad_norm": 0.13020880252619615,
      "learning_rate": 8.506719086493172e-05,
      "loss": 1.139,
      "step": 7787
    },
    {
      "epoch": 0.5613782166798817,
      "grad_norm": 0.24677695335120695,
      "learning_rate": 8.504410589621352e-05,
      "loss": 1.0706,
      "step": 7788
    },
    {
      "epoch": 0.5614502991422187,
      "grad_norm": 0.11108911219031223,
      "learning_rate": 8.502102174272538e-05,
      "loss": 1.0827,
      "step": 7789
    },
    {
      "epoch": 0.5615223816045556,
      "grad_norm": 0.13497475991110505,
      "learning_rate": 8.499793840572552e-05,
      "loss": 1.0011,
      "step": 7790
    },
    {
      "epoch": 0.5615944640668925,
      "grad_norm": 0.11311730692810783,
      "learning_rate": 8.49748558864722e-05,
      "loss": 1.1115,
      "step": 7791
    },
    {
      "epoch": 0.5616665465292294,
      "grad_norm": 0.19175897457035188,
      "learning_rate": 8.49517741862237e-05,
      "loss": 0.978,
      "step": 7792
    },
    {
      "epoch": 0.5617386289915663,
      "grad_norm": 0.1145432738932012,
      "learning_rate": 8.492869330623814e-05,
      "loss": 1.0611,
      "step": 7793
    },
    {
      "epoch": 0.5618107114539033,
      "grad_norm": 0.11828953160217713,
      "learning_rate": 8.490561324777361e-05,
      "loss": 1.0278,
      "step": 7794
    },
    {
      "epoch": 0.5618827939162402,
      "grad_norm": 0.1523734845924617,
      "learning_rate": 8.488253401208816e-05,
      "loss": 0.9873,
      "step": 7795
    },
    {
      "epoch": 0.5619548763785771,
      "grad_norm": 0.13207459607087946,
      "learning_rate": 8.485945560043988e-05,
      "loss": 1.0376,
      "step": 7796
    },
    {
      "epoch": 0.562026958840914,
      "grad_norm": 0.129659554400961,
      "learning_rate": 8.483637801408676e-05,
      "loss": 1.0914,
      "step": 7797
    },
    {
      "epoch": 0.562099041303251,
      "grad_norm": 0.11717195772224513,
      "learning_rate": 8.481330125428666e-05,
      "loss": 1.0268,
      "step": 7798
    },
    {
      "epoch": 0.5621711237655879,
      "grad_norm": 0.12891314427176065,
      "learning_rate": 8.479022532229747e-05,
      "loss": 1.0683,
      "step": 7799
    },
    {
      "epoch": 0.5622432062279248,
      "grad_norm": 0.15757235981090892,
      "learning_rate": 8.476715021937712e-05,
      "loss": 0.9844,
      "step": 7800
    },
    {
      "epoch": 0.5623152886902617,
      "grad_norm": 0.14065751637543705,
      "learning_rate": 8.474407594678337e-05,
      "loss": 1.0876,
      "step": 7801
    },
    {
      "epoch": 0.5623873711525986,
      "grad_norm": 0.3428233198441208,
      "learning_rate": 8.472100250577395e-05,
      "loss": 1.0503,
      "step": 7802
    },
    {
      "epoch": 0.5624594536149354,
      "grad_norm": 0.2172734375079602,
      "learning_rate": 8.469792989760658e-05,
      "loss": 1.1003,
      "step": 7803
    },
    {
      "epoch": 0.5625315360772724,
      "grad_norm": 0.1269279209864412,
      "learning_rate": 8.46748581235389e-05,
      "loss": 0.9784,
      "step": 7804
    },
    {
      "epoch": 0.5626036185396093,
      "grad_norm": 0.16995239514478389,
      "learning_rate": 8.465178718482859e-05,
      "loss": 1.073,
      "step": 7805
    },
    {
      "epoch": 0.5626757010019462,
      "grad_norm": 0.12699831013249357,
      "learning_rate": 8.462871708273316e-05,
      "loss": 1.1117,
      "step": 7806
    },
    {
      "epoch": 0.5627477834642831,
      "grad_norm": 0.140201021688382,
      "learning_rate": 8.460564781851021e-05,
      "loss": 1.0462,
      "step": 7807
    },
    {
      "epoch": 0.56281986592662,
      "grad_norm": 0.11723592852294018,
      "learning_rate": 8.458257939341711e-05,
      "loss": 0.9896,
      "step": 7808
    },
    {
      "epoch": 0.562891948388957,
      "grad_norm": 0.11502136484164505,
      "learning_rate": 8.455951180871141e-05,
      "loss": 1.0885,
      "step": 7809
    },
    {
      "epoch": 0.5629640308512939,
      "grad_norm": 0.1365715075732,
      "learning_rate": 8.453644506565044e-05,
      "loss": 0.9804,
      "step": 7810
    },
    {
      "epoch": 0.5630361133136308,
      "grad_norm": 0.11418464179825506,
      "learning_rate": 8.451337916549157e-05,
      "loss": 1.0058,
      "step": 7811
    },
    {
      "epoch": 0.5631081957759677,
      "grad_norm": 0.12137379661534063,
      "learning_rate": 8.449031410949206e-05,
      "loss": 1.0415,
      "step": 7812
    },
    {
      "epoch": 0.5631802782383046,
      "grad_norm": 0.2489834757043884,
      "learning_rate": 8.446724989890917e-05,
      "loss": 1.1452,
      "step": 7813
    },
    {
      "epoch": 0.5632523607006416,
      "grad_norm": 0.10264525381124287,
      "learning_rate": 8.444418653500014e-05,
      "loss": 1.0917,
      "step": 7814
    },
    {
      "epoch": 0.5633244431629785,
      "grad_norm": 0.1244357565305969,
      "learning_rate": 8.442112401902214e-05,
      "loss": 1.0375,
      "step": 7815
    },
    {
      "epoch": 0.5633965256253154,
      "grad_norm": 0.11586821207903363,
      "learning_rate": 8.439806235223221e-05,
      "loss": 1.058,
      "step": 7816
    },
    {
      "epoch": 0.5634686080876523,
      "grad_norm": 0.14081155588646796,
      "learning_rate": 8.437500153588746e-05,
      "loss": 1.037,
      "step": 7817
    },
    {
      "epoch": 0.5635406905499892,
      "grad_norm": 0.12443013319579631,
      "learning_rate": 8.435194157124492e-05,
      "loss": 1.0659,
      "step": 7818
    },
    {
      "epoch": 0.5636127730123262,
      "grad_norm": 0.2760956523501023,
      "learning_rate": 8.432888245956158e-05,
      "loss": 0.9748,
      "step": 7819
    },
    {
      "epoch": 0.563684855474663,
      "grad_norm": 0.12662144499045436,
      "learning_rate": 8.430582420209432e-05,
      "loss": 1.0165,
      "step": 7820
    },
    {
      "epoch": 0.5637569379369999,
      "grad_norm": 0.13025545005696723,
      "learning_rate": 8.428276680010004e-05,
      "loss": 0.9813,
      "step": 7821
    },
    {
      "epoch": 0.5638290203993368,
      "grad_norm": 0.1876934684491388,
      "learning_rate": 8.425971025483558e-05,
      "loss": 1.0803,
      "step": 7822
    },
    {
      "epoch": 0.5639011028616737,
      "grad_norm": 0.13542223443537366,
      "learning_rate": 8.423665456755776e-05,
      "loss": 1.0626,
      "step": 7823
    },
    {
      "epoch": 0.5639731853240106,
      "grad_norm": 0.17245975559287677,
      "learning_rate": 8.421359973952328e-05,
      "loss": 1.0112,
      "step": 7824
    },
    {
      "epoch": 0.5640452677863476,
      "grad_norm": 0.1605601778984114,
      "learning_rate": 8.419054577198882e-05,
      "loss": 1.115,
      "step": 7825
    },
    {
      "epoch": 0.5641173502486845,
      "grad_norm": 0.13098430424857915,
      "learning_rate": 8.416749266621105e-05,
      "loss": 1.0593,
      "step": 7826
    },
    {
      "epoch": 0.5641894327110214,
      "grad_norm": 0.14267054570962515,
      "learning_rate": 8.414444042344659e-05,
      "loss": 1.0966,
      "step": 7827
    },
    {
      "epoch": 0.5642615151733583,
      "grad_norm": 0.12278134264601609,
      "learning_rate": 8.412138904495196e-05,
      "loss": 1.0497,
      "step": 7828
    },
    {
      "epoch": 0.5643335976356952,
      "grad_norm": 0.12522124438192112,
      "learning_rate": 8.40983385319837e-05,
      "loss": 1.0264,
      "step": 7829
    },
    {
      "epoch": 0.5644056800980322,
      "grad_norm": 0.11964250448396249,
      "learning_rate": 8.407528888579821e-05,
      "loss": 1.034,
      "step": 7830
    },
    {
      "epoch": 0.5644777625603691,
      "grad_norm": 0.11716810102689867,
      "learning_rate": 8.4052240107652e-05,
      "loss": 1.0344,
      "step": 7831
    },
    {
      "epoch": 0.564549845022706,
      "grad_norm": 0.11524394480486261,
      "learning_rate": 8.402919219880134e-05,
      "loss": 0.9377,
      "step": 7832
    },
    {
      "epoch": 0.5646219274850429,
      "grad_norm": 0.15154092570023484,
      "learning_rate": 8.400614516050259e-05,
      "loss": 1.0305,
      "step": 7833
    },
    {
      "epoch": 0.5646940099473798,
      "grad_norm": 0.14165273007565138,
      "learning_rate": 8.398309899401204e-05,
      "loss": 0.9909,
      "step": 7834
    },
    {
      "epoch": 0.5647660924097168,
      "grad_norm": 0.12077505484581086,
      "learning_rate": 8.396005370058584e-05,
      "loss": 1.045,
      "step": 7835
    },
    {
      "epoch": 0.5648381748720537,
      "grad_norm": 0.1282374835376154,
      "learning_rate": 8.393700928148024e-05,
      "loss": 1.0432,
      "step": 7836
    },
    {
      "epoch": 0.5649102573343906,
      "grad_norm": 0.11992174469266716,
      "learning_rate": 8.391396573795133e-05,
      "loss": 1.0769,
      "step": 7837
    },
    {
      "epoch": 0.5649823397967274,
      "grad_norm": 0.1780930442857909,
      "learning_rate": 8.389092307125521e-05,
      "loss": 1.0306,
      "step": 7838
    },
    {
      "epoch": 0.5650544222590643,
      "grad_norm": 0.1316548499245808,
      "learning_rate": 8.386788128264787e-05,
      "loss": 1.0324,
      "step": 7839
    },
    {
      "epoch": 0.5651265047214012,
      "grad_norm": 0.14967961457545906,
      "learning_rate": 8.384484037338533e-05,
      "loss": 1.0897,
      "step": 7840
    },
    {
      "epoch": 0.5651985871837382,
      "grad_norm": 0.12899233394081638,
      "learning_rate": 8.382180034472353e-05,
      "loss": 0.929,
      "step": 7841
    },
    {
      "epoch": 0.5652706696460751,
      "grad_norm": 0.15948931400110236,
      "learning_rate": 8.379876119791836e-05,
      "loss": 1.0751,
      "step": 7842
    },
    {
      "epoch": 0.565342752108412,
      "grad_norm": 0.1222160800003359,
      "learning_rate": 8.37757229342256e-05,
      "loss": 1.012,
      "step": 7843
    },
    {
      "epoch": 0.5654148345707489,
      "grad_norm": 0.12005454582329705,
      "learning_rate": 8.375268555490112e-05,
      "loss": 1.0638,
      "step": 7844
    },
    {
      "epoch": 0.5654869170330858,
      "grad_norm": 0.11172235454521853,
      "learning_rate": 8.372964906120062e-05,
      "loss": 1.084,
      "step": 7845
    },
    {
      "epoch": 0.5655589994954228,
      "grad_norm": 0.11333980627014545,
      "learning_rate": 8.370661345437984e-05,
      "loss": 0.9726,
      "step": 7846
    },
    {
      "epoch": 0.5656310819577597,
      "grad_norm": 0.13339877749544476,
      "learning_rate": 8.368357873569436e-05,
      "loss": 1.0548,
      "step": 7847
    },
    {
      "epoch": 0.5657031644200966,
      "grad_norm": 0.11097486874540755,
      "learning_rate": 8.366054490639978e-05,
      "loss": 1.1561,
      "step": 7848
    },
    {
      "epoch": 0.5657752468824335,
      "grad_norm": 0.12883413578973846,
      "learning_rate": 8.363751196775172e-05,
      "loss": 1.0613,
      "step": 7849
    },
    {
      "epoch": 0.5658473293447704,
      "grad_norm": 0.11267824263755233,
      "learning_rate": 8.361447992100564e-05,
      "loss": 1.008,
      "step": 7850
    },
    {
      "epoch": 0.5659194118071074,
      "grad_norm": 0.173452926695261,
      "learning_rate": 8.359144876741698e-05,
      "loss": 1.0122,
      "step": 7851
    },
    {
      "epoch": 0.5659914942694443,
      "grad_norm": 0.12695744074229257,
      "learning_rate": 8.356841850824115e-05,
      "loss": 1.0554,
      "step": 7852
    },
    {
      "epoch": 0.5660635767317812,
      "grad_norm": 0.11614183089051262,
      "learning_rate": 8.354538914473351e-05,
      "loss": 1.0224,
      "step": 7853
    },
    {
      "epoch": 0.5661356591941181,
      "grad_norm": 0.21864223798945617,
      "learning_rate": 8.35223606781494e-05,
      "loss": 1.1081,
      "step": 7854
    },
    {
      "epoch": 0.5662077416564549,
      "grad_norm": 0.11289759954918935,
      "learning_rate": 8.349933310974402e-05,
      "loss": 0.9801,
      "step": 7855
    },
    {
      "epoch": 0.5662798241187919,
      "grad_norm": 0.1118371932274409,
      "learning_rate": 8.347630644077264e-05,
      "loss": 1.0167,
      "step": 7856
    },
    {
      "epoch": 0.5663519065811288,
      "grad_norm": 0.13877188832235504,
      "learning_rate": 8.345328067249032e-05,
      "loss": 1.0456,
      "step": 7857
    },
    {
      "epoch": 0.5664239890434657,
      "grad_norm": 0.14867563024837757,
      "learning_rate": 8.34302558061523e-05,
      "loss": 1.0605,
      "step": 7858
    },
    {
      "epoch": 0.5664960715058026,
      "grad_norm": 0.14729178176132052,
      "learning_rate": 8.340723184301353e-05,
      "loss": 1.0192,
      "step": 7859
    },
    {
      "epoch": 0.5665681539681395,
      "grad_norm": 0.1603118611486046,
      "learning_rate": 8.338420878432911e-05,
      "loss": 1.0258,
      "step": 7860
    },
    {
      "epoch": 0.5666402364304765,
      "grad_norm": 0.12406300203778754,
      "learning_rate": 8.336118663135389e-05,
      "loss": 1.0974,
      "step": 7861
    },
    {
      "epoch": 0.5667123188928134,
      "grad_norm": 0.13783463972935406,
      "learning_rate": 8.333816538534292e-05,
      "loss": 1.1464,
      "step": 7862
    },
    {
      "epoch": 0.5667844013551503,
      "grad_norm": 0.1452682409759059,
      "learning_rate": 8.331514504755098e-05,
      "loss": 1.0545,
      "step": 7863
    },
    {
      "epoch": 0.5668564838174872,
      "grad_norm": 0.1278826157483398,
      "learning_rate": 8.329212561923291e-05,
      "loss": 1.0989,
      "step": 7864
    },
    {
      "epoch": 0.5669285662798241,
      "grad_norm": 0.13115311342571292,
      "learning_rate": 8.326910710164342e-05,
      "loss": 1.108,
      "step": 7865
    },
    {
      "epoch": 0.567000648742161,
      "grad_norm": 0.12808731436771809,
      "learning_rate": 8.324608949603733e-05,
      "loss": 1.0173,
      "step": 7866
    },
    {
      "epoch": 0.567072731204498,
      "grad_norm": 0.11132394440865309,
      "learning_rate": 8.322307280366922e-05,
      "loss": 1.043,
      "step": 7867
    },
    {
      "epoch": 0.5671448136668349,
      "grad_norm": 0.11131640802256337,
      "learning_rate": 8.320005702579377e-05,
      "loss": 1.0849,
      "step": 7868
    },
    {
      "epoch": 0.5672168961291718,
      "grad_norm": 0.12647790635120615,
      "learning_rate": 8.317704216366547e-05,
      "loss": 0.9979,
      "step": 7869
    },
    {
      "epoch": 0.5672889785915087,
      "grad_norm": 0.12500381750222056,
      "learning_rate": 8.315402821853886e-05,
      "loss": 1.0646,
      "step": 7870
    },
    {
      "epoch": 0.5673610610538457,
      "grad_norm": 0.1169908948056688,
      "learning_rate": 8.313101519166844e-05,
      "loss": 1.0451,
      "step": 7871
    },
    {
      "epoch": 0.5674331435161825,
      "grad_norm": 0.18572904867725937,
      "learning_rate": 8.310800308430861e-05,
      "loss": 1.0443,
      "step": 7872
    },
    {
      "epoch": 0.5675052259785194,
      "grad_norm": 0.1212497729727648,
      "learning_rate": 8.308499189771374e-05,
      "loss": 1.0198,
      "step": 7873
    },
    {
      "epoch": 0.5675773084408563,
      "grad_norm": 0.1293468495751288,
      "learning_rate": 8.30619816331381e-05,
      "loss": 1.004,
      "step": 7874
    },
    {
      "epoch": 0.5676493909031932,
      "grad_norm": 0.11780240350981627,
      "learning_rate": 8.303897229183602e-05,
      "loss": 1.0517,
      "step": 7875
    },
    {
      "epoch": 0.5677214733655301,
      "grad_norm": 0.1477466218999907,
      "learning_rate": 8.30159638750617e-05,
      "loss": 1.0102,
      "step": 7876
    },
    {
      "epoch": 0.5677935558278671,
      "grad_norm": 0.11199837259428944,
      "learning_rate": 8.299295638406929e-05,
      "loss": 1.1363,
      "step": 7877
    },
    {
      "epoch": 0.567865638290204,
      "grad_norm": 0.09897639510489797,
      "learning_rate": 8.296994982011291e-05,
      "loss": 1.0365,
      "step": 7878
    },
    {
      "epoch": 0.5679377207525409,
      "grad_norm": 0.13844921463529036,
      "learning_rate": 8.294694418444657e-05,
      "loss": 0.9939,
      "step": 7879
    },
    {
      "epoch": 0.5680098032148778,
      "grad_norm": 0.11416008732372883,
      "learning_rate": 8.292393947832439e-05,
      "loss": 1.0248,
      "step": 7880
    },
    {
      "epoch": 0.5680818856772147,
      "grad_norm": 0.11303342102492944,
      "learning_rate": 8.290093570300025e-05,
      "loss": 0.9446,
      "step": 7881
    },
    {
      "epoch": 0.5681539681395517,
      "grad_norm": 0.13091324776137717,
      "learning_rate": 8.287793285972812e-05,
      "loss": 1.0341,
      "step": 7882
    },
    {
      "epoch": 0.5682260506018886,
      "grad_norm": 0.11776260324237894,
      "learning_rate": 8.285493094976174e-05,
      "loss": 1.0341,
      "step": 7883
    },
    {
      "epoch": 0.5682981330642255,
      "grad_norm": 0.10646555707020001,
      "learning_rate": 8.28319299743551e-05,
      "loss": 0.9944,
      "step": 7884
    },
    {
      "epoch": 0.5683702155265624,
      "grad_norm": 0.43065762306963606,
      "learning_rate": 8.280892993476183e-05,
      "loss": 0.9879,
      "step": 7885
    },
    {
      "epoch": 0.5684422979888993,
      "grad_norm": 0.1054296595102709,
      "learning_rate": 8.278593083223569e-05,
      "loss": 1.1022,
      "step": 7886
    },
    {
      "epoch": 0.5685143804512363,
      "grad_norm": 0.1116640079427072,
      "learning_rate": 8.276293266803027e-05,
      "loss": 1.0296,
      "step": 7887
    },
    {
      "epoch": 0.5685864629135732,
      "grad_norm": 1.0862243817350064,
      "learning_rate": 8.273993544339929e-05,
      "loss": 0.9919,
      "step": 7888
    },
    {
      "epoch": 0.5686585453759101,
      "grad_norm": 0.12304590472838668,
      "learning_rate": 8.271693915959621e-05,
      "loss": 1.1183,
      "step": 7889
    },
    {
      "epoch": 0.5687306278382469,
      "grad_norm": 0.11822125928236389,
      "learning_rate": 8.269394381787459e-05,
      "loss": 1.1016,
      "step": 7890
    },
    {
      "epoch": 0.5688027103005838,
      "grad_norm": 0.13022949226863406,
      "learning_rate": 8.267094941948783e-05,
      "loss": 0.9327,
      "step": 7891
    },
    {
      "epoch": 0.5688747927629207,
      "grad_norm": 0.15304949175524724,
      "learning_rate": 8.264795596568935e-05,
      "loss": 1.0718,
      "step": 7892
    },
    {
      "epoch": 0.5689468752252577,
      "grad_norm": 0.13171498695413564,
      "learning_rate": 8.262496345773251e-05,
      "loss": 1.0431,
      "step": 7893
    },
    {
      "epoch": 0.5690189576875946,
      "grad_norm": 0.1209936569941994,
      "learning_rate": 8.260197189687063e-05,
      "loss": 1.0316,
      "step": 7894
    },
    {
      "epoch": 0.5690910401499315,
      "grad_norm": 0.13905385581230775,
      "learning_rate": 8.257898128435692e-05,
      "loss": 1.0111,
      "step": 7895
    },
    {
      "epoch": 0.5691631226122684,
      "grad_norm": 0.11888651848663803,
      "learning_rate": 8.255599162144456e-05,
      "loss": 1.0235,
      "step": 7896
    },
    {
      "epoch": 0.5692352050746053,
      "grad_norm": 0.12475787044642056,
      "learning_rate": 8.253300290938671e-05,
      "loss": 1.0133,
      "step": 7897
    },
    {
      "epoch": 0.5693072875369423,
      "grad_norm": 0.1113763695050268,
      "learning_rate": 8.251001514943651e-05,
      "loss": 1.0205,
      "step": 7898
    },
    {
      "epoch": 0.5693793699992792,
      "grad_norm": 0.12293276730191433,
      "learning_rate": 8.248702834284693e-05,
      "loss": 1.0752,
      "step": 7899
    },
    {
      "epoch": 0.5694514524616161,
      "grad_norm": 0.1247143524835507,
      "learning_rate": 8.246404249087097e-05,
      "loss": 1.0715,
      "step": 7900
    },
    {
      "epoch": 0.569523534923953,
      "grad_norm": 0.13017040108336125,
      "learning_rate": 8.244105759476157e-05,
      "loss": 1.0819,
      "step": 7901
    },
    {
      "epoch": 0.5695956173862899,
      "grad_norm": 0.14054270111300807,
      "learning_rate": 8.241807365577165e-05,
      "loss": 1.0428,
      "step": 7902
    },
    {
      "epoch": 0.5696676998486269,
      "grad_norm": 0.11764314263246901,
      "learning_rate": 8.239509067515399e-05,
      "loss": 1.057,
      "step": 7903
    },
    {
      "epoch": 0.5697397823109638,
      "grad_norm": 0.13191673918245136,
      "learning_rate": 8.237210865416138e-05,
      "loss": 1.0679,
      "step": 7904
    },
    {
      "epoch": 0.5698118647733007,
      "grad_norm": 0.18354333368984763,
      "learning_rate": 8.234912759404656e-05,
      "loss": 1.0191,
      "step": 7905
    },
    {
      "epoch": 0.5698839472356376,
      "grad_norm": 0.10620866813235301,
      "learning_rate": 8.23261474960622e-05,
      "loss": 1.0715,
      "step": 7906
    },
    {
      "epoch": 0.5699560296979744,
      "grad_norm": 0.11409047380158434,
      "learning_rate": 8.230316836146091e-05,
      "loss": 1.0866,
      "step": 7907
    },
    {
      "epoch": 0.5700281121603114,
      "grad_norm": 0.12940158128995458,
      "learning_rate": 8.228019019149526e-05,
      "loss": 1.0677,
      "step": 7908
    },
    {
      "epoch": 0.5701001946226483,
      "grad_norm": 0.1350305206292877,
      "learning_rate": 8.225721298741776e-05,
      "loss": 0.9944,
      "step": 7909
    },
    {
      "epoch": 0.5701722770849852,
      "grad_norm": 0.1225468633059851,
      "learning_rate": 8.223423675048091e-05,
      "loss": 1.0393,
      "step": 7910
    },
    {
      "epoch": 0.5702443595473221,
      "grad_norm": 0.13396977131423232,
      "learning_rate": 8.22112614819371e-05,
      "loss": 0.9976,
      "step": 7911
    },
    {
      "epoch": 0.570316442009659,
      "grad_norm": 0.12451337819904115,
      "learning_rate": 8.218828718303865e-05,
      "loss": 1.2405,
      "step": 7912
    },
    {
      "epoch": 0.570388524471996,
      "grad_norm": 0.13055214546782876,
      "learning_rate": 8.216531385503794e-05,
      "loss": 1.045,
      "step": 7913
    },
    {
      "epoch": 0.5704606069343329,
      "grad_norm": 0.12046614257097893,
      "learning_rate": 8.214234149918714e-05,
      "loss": 1.116,
      "step": 7914
    },
    {
      "epoch": 0.5705326893966698,
      "grad_norm": 0.11939601896646762,
      "learning_rate": 8.211937011673849e-05,
      "loss": 1.1016,
      "step": 7915
    },
    {
      "epoch": 0.5706047718590067,
      "grad_norm": 0.10729040271694348,
      "learning_rate": 8.209639970894413e-05,
      "loss": 1.0372,
      "step": 7916
    },
    {
      "epoch": 0.5706768543213436,
      "grad_norm": 0.11941643363189507,
      "learning_rate": 8.207343027705621e-05,
      "loss": 1.088,
      "step": 7917
    },
    {
      "epoch": 0.5707489367836805,
      "grad_norm": 0.11570390189245255,
      "learning_rate": 8.205046182232664e-05,
      "loss": 1.093,
      "step": 7918
    },
    {
      "epoch": 0.5708210192460175,
      "grad_norm": 0.12497328068378451,
      "learning_rate": 8.202749434600752e-05,
      "loss": 1.0638,
      "step": 7919
    },
    {
      "epoch": 0.5708931017083544,
      "grad_norm": 0.157130248497148,
      "learning_rate": 8.200452784935073e-05,
      "loss": 1.027,
      "step": 7920
    },
    {
      "epoch": 0.5709651841706913,
      "grad_norm": 0.12269085288490464,
      "learning_rate": 8.19815623336082e-05,
      "loss": 1.0784,
      "step": 7921
    },
    {
      "epoch": 0.5710372666330282,
      "grad_norm": 0.11368172985735057,
      "learning_rate": 8.19585978000317e-05,
      "loss": 1.0407,
      "step": 7922
    },
    {
      "epoch": 0.5711093490953651,
      "grad_norm": 0.1318303515720093,
      "learning_rate": 8.193563424987298e-05,
      "loss": 1.0048,
      "step": 7923
    },
    {
      "epoch": 0.571181431557702,
      "grad_norm": 0.12682581840114485,
      "learning_rate": 8.191267168438385e-05,
      "loss": 0.9904,
      "step": 7924
    },
    {
      "epoch": 0.5712535140200389,
      "grad_norm": 0.1176755384223193,
      "learning_rate": 8.188971010481593e-05,
      "loss": 0.9951,
      "step": 7925
    },
    {
      "epoch": 0.5713255964823758,
      "grad_norm": 0.12800687729528085,
      "learning_rate": 8.186674951242082e-05,
      "loss": 1.0191,
      "step": 7926
    },
    {
      "epoch": 0.5713976789447127,
      "grad_norm": 0.19450239094573885,
      "learning_rate": 8.184378990845005e-05,
      "loss": 1.025,
      "step": 7927
    },
    {
      "epoch": 0.5714697614070496,
      "grad_norm": 0.11494538304212286,
      "learning_rate": 8.18208312941552e-05,
      "loss": 1.0636,
      "step": 7928
    },
    {
      "epoch": 0.5715418438693866,
      "grad_norm": 0.1241843644512995,
      "learning_rate": 8.17978736707877e-05,
      "loss": 1.1128,
      "step": 7929
    },
    {
      "epoch": 0.5716139263317235,
      "grad_norm": 0.11710128895183737,
      "learning_rate": 8.177491703959889e-05,
      "loss": 1.0703,
      "step": 7930
    },
    {
      "epoch": 0.5716860087940604,
      "grad_norm": 0.12661030268505202,
      "learning_rate": 8.175196140184012e-05,
      "loss": 1.0359,
      "step": 7931
    },
    {
      "epoch": 0.5717580912563973,
      "grad_norm": 0.1082914300191461,
      "learning_rate": 8.172900675876276e-05,
      "loss": 1.0162,
      "step": 7932
    },
    {
      "epoch": 0.5718301737187342,
      "grad_norm": 0.10777755380959068,
      "learning_rate": 8.170605311161798e-05,
      "loss": 1.0716,
      "step": 7933
    },
    {
      "epoch": 0.5719022561810712,
      "grad_norm": 0.11163192008919941,
      "learning_rate": 8.168310046165695e-05,
      "loss": 0.9721,
      "step": 7934
    },
    {
      "epoch": 0.5719743386434081,
      "grad_norm": 0.10911827999843372,
      "learning_rate": 8.166014881013084e-05,
      "loss": 1.0451,
      "step": 7935
    },
    {
      "epoch": 0.572046421105745,
      "grad_norm": 0.12303672057850992,
      "learning_rate": 8.163719815829063e-05,
      "loss": 1.0515,
      "step": 7936
    },
    {
      "epoch": 0.5721185035680819,
      "grad_norm": 0.11853548707977975,
      "learning_rate": 8.161424850738746e-05,
      "loss": 1.0166,
      "step": 7937
    },
    {
      "epoch": 0.5721905860304188,
      "grad_norm": 0.11960092485471246,
      "learning_rate": 8.15912998586722e-05,
      "loss": 1.0014,
      "step": 7938
    },
    {
      "epoch": 0.5722626684927558,
      "grad_norm": 0.11296200299703539,
      "learning_rate": 8.156835221339582e-05,
      "loss": 1.0843,
      "step": 7939
    },
    {
      "epoch": 0.5723347509550927,
      "grad_norm": 0.13175792127373243,
      "learning_rate": 8.154540557280908e-05,
      "loss": 0.9297,
      "step": 7940
    },
    {
      "epoch": 0.5724068334174296,
      "grad_norm": 0.13616173321723737,
      "learning_rate": 8.15224599381629e-05,
      "loss": 1.0543,
      "step": 7941
    },
    {
      "epoch": 0.5724789158797664,
      "grad_norm": 0.10997761889965058,
      "learning_rate": 8.149951531070792e-05,
      "loss": 1.0225,
      "step": 7942
    },
    {
      "epoch": 0.5725509983421033,
      "grad_norm": 0.14803983633903817,
      "learning_rate": 8.14765716916949e-05,
      "loss": 1.026,
      "step": 7943
    },
    {
      "epoch": 0.5726230808044402,
      "grad_norm": 0.18809442889370312,
      "learning_rate": 8.145362908237442e-05,
      "loss": 1.0467,
      "step": 7944
    },
    {
      "epoch": 0.5726951632667772,
      "grad_norm": 0.2838253345962623,
      "learning_rate": 8.143068748399704e-05,
      "loss": 0.9948,
      "step": 7945
    },
    {
      "epoch": 0.5727672457291141,
      "grad_norm": 0.13032160199853182,
      "learning_rate": 8.140774689781335e-05,
      "loss": 1.013,
      "step": 7946
    },
    {
      "epoch": 0.572839328191451,
      "grad_norm": 0.22054999896839625,
      "learning_rate": 8.13848073250738e-05,
      "loss": 1.0106,
      "step": 7947
    },
    {
      "epoch": 0.5729114106537879,
      "grad_norm": 0.12846941931025302,
      "learning_rate": 8.136186876702876e-05,
      "loss": 1.0136,
      "step": 7948
    },
    {
      "epoch": 0.5729834931161248,
      "grad_norm": 0.12464482978154891,
      "learning_rate": 8.13389312249286e-05,
      "loss": 1.0718,
      "step": 7949
    },
    {
      "epoch": 0.5730555755784618,
      "grad_norm": 0.13540539504543297,
      "learning_rate": 8.131599470002367e-05,
      "loss": 0.9417,
      "step": 7950
    },
    {
      "epoch": 0.5731276580407987,
      "grad_norm": 0.148347889620123,
      "learning_rate": 8.129305919356418e-05,
      "loss": 0.9478,
      "step": 7951
    },
    {
      "epoch": 0.5731997405031356,
      "grad_norm": 0.10991753115545154,
      "learning_rate": 8.12701247068003e-05,
      "loss": 1.1267,
      "step": 7952
    },
    {
      "epoch": 0.5732718229654725,
      "grad_norm": 0.12651226807473348,
      "learning_rate": 8.124719124098217e-05,
      "loss": 1.0117,
      "step": 7953
    },
    {
      "epoch": 0.5733439054278094,
      "grad_norm": 0.1515962771137772,
      "learning_rate": 8.122425879735992e-05,
      "loss": 1.0393,
      "step": 7954
    },
    {
      "epoch": 0.5734159878901464,
      "grad_norm": 0.1400960646778876,
      "learning_rate": 8.120132737718356e-05,
      "loss": 1.2103,
      "step": 7955
    },
    {
      "epoch": 0.5734880703524833,
      "grad_norm": 0.3837837913930536,
      "learning_rate": 8.117839698170299e-05,
      "loss": 1.0389,
      "step": 7956
    },
    {
      "epoch": 0.5735601528148202,
      "grad_norm": 0.15481236571082296,
      "learning_rate": 8.115546761216822e-05,
      "loss": 1.0682,
      "step": 7957
    },
    {
      "epoch": 0.5736322352771571,
      "grad_norm": 0.10936735902755956,
      "learning_rate": 8.113253926982896e-05,
      "loss": 1.1071,
      "step": 7958
    },
    {
      "epoch": 0.5737043177394939,
      "grad_norm": 0.1498770359208209,
      "learning_rate": 8.110961195593517e-05,
      "loss": 1.0284,
      "step": 7959
    },
    {
      "epoch": 0.5737764002018308,
      "grad_norm": 0.10634334194342726,
      "learning_rate": 8.108668567173651e-05,
      "loss": 1.1196,
      "step": 7960
    },
    {
      "epoch": 0.5738484826641678,
      "grad_norm": 0.11612849254268019,
      "learning_rate": 8.106376041848272e-05,
      "loss": 0.982,
      "step": 7961
    },
    {
      "epoch": 0.5739205651265047,
      "grad_norm": 0.11850111253098801,
      "learning_rate": 8.104083619742331e-05,
      "loss": 1.0629,
      "step": 7962
    },
    {
      "epoch": 0.5739926475888416,
      "grad_norm": 0.12210438020474208,
      "learning_rate": 8.101791300980802e-05,
      "loss": 1.0229,
      "step": 7963
    },
    {
      "epoch": 0.5740647300511785,
      "grad_norm": 0.13548688386984584,
      "learning_rate": 8.099499085688626e-05,
      "loss": 1.0916,
      "step": 7964
    },
    {
      "epoch": 0.5741368125135154,
      "grad_norm": 0.16988579619663927,
      "learning_rate": 8.097206973990756e-05,
      "loss": 1.1361,
      "step": 7965
    },
    {
      "epoch": 0.5742088949758524,
      "grad_norm": 0.13534524504475515,
      "learning_rate": 8.094914966012124e-05,
      "loss": 0.996,
      "step": 7966
    },
    {
      "epoch": 0.5742809774381893,
      "grad_norm": 0.12387439738782456,
      "learning_rate": 8.092623061877668e-05,
      "loss": 1.0486,
      "step": 7967
    },
    {
      "epoch": 0.5743530599005262,
      "grad_norm": 0.12417207929793507,
      "learning_rate": 8.090331261712322e-05,
      "loss": 1.0746,
      "step": 7968
    },
    {
      "epoch": 0.5744251423628631,
      "grad_norm": 0.11445804029431808,
      "learning_rate": 8.08803956564101e-05,
      "loss": 1.0309,
      "step": 7969
    },
    {
      "epoch": 0.5744972248252,
      "grad_norm": 0.13005253515856632,
      "learning_rate": 8.085747973788644e-05,
      "loss": 1.0065,
      "step": 7970
    },
    {
      "epoch": 0.574569307287537,
      "grad_norm": 0.1949054101656449,
      "learning_rate": 8.083456486280136e-05,
      "loss": 1.1089,
      "step": 7971
    },
    {
      "epoch": 0.5746413897498739,
      "grad_norm": 0.10016864867510213,
      "learning_rate": 8.081165103240397e-05,
      "loss": 1.0198,
      "step": 7972
    },
    {
      "epoch": 0.5747134722122108,
      "grad_norm": 0.11724865545338257,
      "learning_rate": 8.078873824794329e-05,
      "loss": 1.0554,
      "step": 7973
    },
    {
      "epoch": 0.5747855546745477,
      "grad_norm": 0.13490333919223074,
      "learning_rate": 8.076582651066822e-05,
      "loss": 1.0841,
      "step": 7974
    },
    {
      "epoch": 0.5748576371368846,
      "grad_norm": 0.11871845504156321,
      "learning_rate": 8.074291582182766e-05,
      "loss": 1.0745,
      "step": 7975
    },
    {
      "epoch": 0.5749297195992215,
      "grad_norm": 0.13329789256038055,
      "learning_rate": 8.072000618267048e-05,
      "loss": 1.008,
      "step": 7976
    },
    {
      "epoch": 0.5750018020615584,
      "grad_norm": 0.14500447287887885,
      "learning_rate": 8.069709759444548e-05,
      "loss": 1.0754,
      "step": 7977
    },
    {
      "epoch": 0.5750738845238953,
      "grad_norm": 0.1263144232294656,
      "learning_rate": 8.067419005840132e-05,
      "loss": 1.0381,
      "step": 7978
    },
    {
      "epoch": 0.5751459669862322,
      "grad_norm": 0.15164437740510853,
      "learning_rate": 8.065128357578671e-05,
      "loss": 1.0422,
      "step": 7979
    },
    {
      "epoch": 0.5752180494485691,
      "grad_norm": 0.12277180802372155,
      "learning_rate": 8.06283781478502e-05,
      "loss": 1.0488,
      "step": 7980
    },
    {
      "epoch": 0.575290131910906,
      "grad_norm": 0.13430705806075516,
      "learning_rate": 8.060547377584045e-05,
      "loss": 1.088,
      "step": 7981
    },
    {
      "epoch": 0.575362214373243,
      "grad_norm": 0.21238450919365326,
      "learning_rate": 8.058257046100585e-05,
      "loss": 1.0967,
      "step": 7982
    },
    {
      "epoch": 0.5754342968355799,
      "grad_norm": 0.16640251633158448,
      "learning_rate": 8.055966820459487e-05,
      "loss": 1.0492,
      "step": 7983
    },
    {
      "epoch": 0.5755063792979168,
      "grad_norm": 0.15052270228058698,
      "learning_rate": 8.053676700785589e-05,
      "loss": 1.0055,
      "step": 7984
    },
    {
      "epoch": 0.5755784617602537,
      "grad_norm": 0.12329939251961994,
      "learning_rate": 8.051386687203725e-05,
      "loss": 1.0708,
      "step": 7985
    },
    {
      "epoch": 0.5756505442225907,
      "grad_norm": 0.14264545525472785,
      "learning_rate": 8.049096779838719e-05,
      "loss": 0.9653,
      "step": 7986
    },
    {
      "epoch": 0.5757226266849276,
      "grad_norm": 0.12056113784878857,
      "learning_rate": 8.046806978815391e-05,
      "loss": 1.0294,
      "step": 7987
    },
    {
      "epoch": 0.5757947091472645,
      "grad_norm": 0.1169469807226169,
      "learning_rate": 8.044517284258559e-05,
      "loss": 1.1882,
      "step": 7988
    },
    {
      "epoch": 0.5758667916096014,
      "grad_norm": 0.12281794579528048,
      "learning_rate": 8.042227696293023e-05,
      "loss": 1.0446,
      "step": 7989
    },
    {
      "epoch": 0.5759388740719383,
      "grad_norm": 0.12221345241816879,
      "learning_rate": 8.039938215043597e-05,
      "loss": 1.0441,
      "step": 7990
    },
    {
      "epoch": 0.5760109565342753,
      "grad_norm": 0.1336927562184419,
      "learning_rate": 8.037648840635073e-05,
      "loss": 1.0145,
      "step": 7991
    },
    {
      "epoch": 0.5760830389966122,
      "grad_norm": 0.12677281619202535,
      "learning_rate": 8.035359573192243e-05,
      "loss": 1.0429,
      "step": 7992
    },
    {
      "epoch": 0.576155121458949,
      "grad_norm": 0.13106814334954636,
      "learning_rate": 8.033070412839891e-05,
      "loss": 1.0173,
      "step": 7993
    },
    {
      "epoch": 0.5762272039212859,
      "grad_norm": 0.11990081585865212,
      "learning_rate": 8.0307813597028e-05,
      "loss": 1.048,
      "step": 7994
    },
    {
      "epoch": 0.5762992863836228,
      "grad_norm": 0.12471383201900836,
      "learning_rate": 8.02849241390574e-05,
      "loss": 0.9557,
      "step": 7995
    },
    {
      "epoch": 0.5763713688459597,
      "grad_norm": 0.1292931203879313,
      "learning_rate": 8.026203575573484e-05,
      "loss": 0.9989,
      "step": 7996
    },
    {
      "epoch": 0.5764434513082967,
      "grad_norm": 0.1211093107937822,
      "learning_rate": 8.023914844830789e-05,
      "loss": 1.1404,
      "step": 7997
    },
    {
      "epoch": 0.5765155337706336,
      "grad_norm": 0.13842752612136464,
      "learning_rate": 8.021626221802411e-05,
      "loss": 0.9509,
      "step": 7998
    },
    {
      "epoch": 0.5765876162329705,
      "grad_norm": 0.11289390263251452,
      "learning_rate": 8.019337706613107e-05,
      "loss": 1.0615,
      "step": 7999
    },
    {
      "epoch": 0.5766596986953074,
      "grad_norm": 0.12512201685228277,
      "learning_rate": 8.017049299387618e-05,
      "loss": 1.0684,
      "step": 8000
    },
    {
      "epoch": 0.5767317811576443,
      "grad_norm": 0.24281022585867226,
      "learning_rate": 8.014761000250682e-05,
      "loss": 1.0254,
      "step": 8001
    },
    {
      "epoch": 0.5768038636199813,
      "grad_norm": 0.15366035979065126,
      "learning_rate": 8.012472809327028e-05,
      "loss": 1.0971,
      "step": 8002
    },
    {
      "epoch": 0.5768759460823182,
      "grad_norm": 0.11232003720426126,
      "learning_rate": 8.010184726741387e-05,
      "loss": 1.06,
      "step": 8003
    },
    {
      "epoch": 0.5769480285446551,
      "grad_norm": 0.11718744379260136,
      "learning_rate": 8.007896752618486e-05,
      "loss": 1.0217,
      "step": 8004
    },
    {
      "epoch": 0.577020111006992,
      "grad_norm": 0.13489608913921794,
      "learning_rate": 8.00560888708303e-05,
      "loss": 1.052,
      "step": 8005
    },
    {
      "epoch": 0.5770921934693289,
      "grad_norm": 0.15746666094497375,
      "learning_rate": 8.003321130259728e-05,
      "loss": 1.0602,
      "step": 8006
    },
    {
      "epoch": 0.5771642759316659,
      "grad_norm": 0.2000827725483357,
      "learning_rate": 8.001033482273292e-05,
      "loss": 1.04,
      "step": 8007
    },
    {
      "epoch": 0.5772363583940028,
      "grad_norm": 0.16963002811524858,
      "learning_rate": 7.998745943248415e-05,
      "loss": 0.9727,
      "step": 8008
    },
    {
      "epoch": 0.5773084408563397,
      "grad_norm": 0.11117123227659748,
      "learning_rate": 7.996458513309787e-05,
      "loss": 0.9996,
      "step": 8009
    },
    {
      "epoch": 0.5773805233186766,
      "grad_norm": 0.12602483774989115,
      "learning_rate": 7.994171192582096e-05,
      "loss": 1.11,
      "step": 8010
    },
    {
      "epoch": 0.5774526057810134,
      "grad_norm": 0.1525612214708458,
      "learning_rate": 7.991883981190012e-05,
      "loss": 1.0288,
      "step": 8011
    },
    {
      "epoch": 0.5775246882433503,
      "grad_norm": 0.1387958403474128,
      "learning_rate": 7.989596879258224e-05,
      "loss": 0.999,
      "step": 8012
    },
    {
      "epoch": 0.5775967707056873,
      "grad_norm": 0.14353284568965963,
      "learning_rate": 7.987309886911389e-05,
      "loss": 1.0623,
      "step": 8013
    },
    {
      "epoch": 0.5776688531680242,
      "grad_norm": 0.14291670609683546,
      "learning_rate": 7.985023004274174e-05,
      "loss": 1.0754,
      "step": 8014
    },
    {
      "epoch": 0.5777409356303611,
      "grad_norm": 0.15209385935433156,
      "learning_rate": 7.982736231471224e-05,
      "loss": 1.0325,
      "step": 8015
    },
    {
      "epoch": 0.577813018092698,
      "grad_norm": 0.19629099083260645,
      "learning_rate": 7.980449568627204e-05,
      "loss": 1.077,
      "step": 8016
    },
    {
      "epoch": 0.577885100555035,
      "grad_norm": 0.12277194920544159,
      "learning_rate": 7.978163015866748e-05,
      "loss": 0.9999,
      "step": 8017
    },
    {
      "epoch": 0.5779571830173719,
      "grad_norm": 0.12658761951051195,
      "learning_rate": 7.975876573314496e-05,
      "loss": 1.0445,
      "step": 8018
    },
    {
      "epoch": 0.5780292654797088,
      "grad_norm": 0.1250031867890307,
      "learning_rate": 7.973590241095078e-05,
      "loss": 1.0255,
      "step": 8019
    },
    {
      "epoch": 0.5781013479420457,
      "grad_norm": 0.11578395242472861,
      "learning_rate": 7.97130401933312e-05,
      "loss": 1.0603,
      "step": 8020
    },
    {
      "epoch": 0.5781734304043826,
      "grad_norm": 0.11126175824410071,
      "learning_rate": 7.969017908153242e-05,
      "loss": 1.0907,
      "step": 8021
    },
    {
      "epoch": 0.5782455128667195,
      "grad_norm": 1.5501094623826759,
      "learning_rate": 7.96673190768006e-05,
      "loss": 1.0587,
      "step": 8022
    },
    {
      "epoch": 0.5783175953290565,
      "grad_norm": 0.12310052336178332,
      "learning_rate": 7.964446018038177e-05,
      "loss": 1.0391,
      "step": 8023
    },
    {
      "epoch": 0.5783896777913934,
      "grad_norm": 0.11313373536181867,
      "learning_rate": 7.962160239352195e-05,
      "loss": 1.0203,
      "step": 8024
    },
    {
      "epoch": 0.5784617602537303,
      "grad_norm": 0.15508435943783236,
      "learning_rate": 7.959874571746713e-05,
      "loss": 1.1103,
      "step": 8025
    },
    {
      "epoch": 0.5785338427160672,
      "grad_norm": 0.1348812369779916,
      "learning_rate": 7.957589015346321e-05,
      "loss": 1.0719,
      "step": 8026
    },
    {
      "epoch": 0.5786059251784041,
      "grad_norm": 0.7612188591891429,
      "learning_rate": 7.955303570275598e-05,
      "loss": 1.1119,
      "step": 8027
    },
    {
      "epoch": 0.578678007640741,
      "grad_norm": 0.1459815243066931,
      "learning_rate": 7.95301823665912e-05,
      "loss": 0.9575,
      "step": 8028
    },
    {
      "epoch": 0.5787500901030779,
      "grad_norm": 0.1132617171115689,
      "learning_rate": 7.95073301462146e-05,
      "loss": 1.0471,
      "step": 8029
    },
    {
      "epoch": 0.5788221725654148,
      "grad_norm": 0.11835310827092572,
      "learning_rate": 7.948447904287191e-05,
      "loss": 1.036,
      "step": 8030
    },
    {
      "epoch": 0.5788942550277517,
      "grad_norm": 0.18098341349574612,
      "learning_rate": 7.94616290578086e-05,
      "loss": 1.1034,
      "step": 8031
    },
    {
      "epoch": 0.5789663374900886,
      "grad_norm": 0.1218393821529151,
      "learning_rate": 7.943878019227028e-05,
      "loss": 1.0401,
      "step": 8032
    },
    {
      "epoch": 0.5790384199524256,
      "grad_norm": 0.14449339062925973,
      "learning_rate": 7.941593244750232e-05,
      "loss": 1.0944,
      "step": 8033
    },
    {
      "epoch": 0.5791105024147625,
      "grad_norm": 0.33725138152721307,
      "learning_rate": 7.939308582475025e-05,
      "loss": 1.0896,
      "step": 8034
    },
    {
      "epoch": 0.5791825848770994,
      "grad_norm": 0.12618063916992583,
      "learning_rate": 7.937024032525933e-05,
      "loss": 1.0489,
      "step": 8035
    },
    {
      "epoch": 0.5792546673394363,
      "grad_norm": 0.11818739408280354,
      "learning_rate": 7.934739595027491e-05,
      "loss": 1.0489,
      "step": 8036
    },
    {
      "epoch": 0.5793267498017732,
      "grad_norm": 0.26828042258786505,
      "learning_rate": 7.93245527010421e-05,
      "loss": 0.9608,
      "step": 8037
    },
    {
      "epoch": 0.5793988322641102,
      "grad_norm": 0.12897996030757392,
      "learning_rate": 7.93017105788062e-05,
      "loss": 1.0328,
      "step": 8038
    },
    {
      "epoch": 0.5794709147264471,
      "grad_norm": 0.1104022993083663,
      "learning_rate": 7.927886958481223e-05,
      "loss": 1.0223,
      "step": 8039
    },
    {
      "epoch": 0.579542997188784,
      "grad_norm": 0.11576665978621321,
      "learning_rate": 7.925602972030525e-05,
      "loss": 0.9995,
      "step": 8040
    },
    {
      "epoch": 0.5796150796511209,
      "grad_norm": 0.15324627011250605,
      "learning_rate": 7.923319098653024e-05,
      "loss": 1.1108,
      "step": 8041
    },
    {
      "epoch": 0.5796871621134578,
      "grad_norm": 0.12828413608426442,
      "learning_rate": 7.921035338473206e-05,
      "loss": 0.9692,
      "step": 8042
    },
    {
      "epoch": 0.5797592445757948,
      "grad_norm": 0.11660065630492553,
      "learning_rate": 7.918751691615563e-05,
      "loss": 0.9889,
      "step": 8043
    },
    {
      "epoch": 0.5798313270381317,
      "grad_norm": 0.20408022259344538,
      "learning_rate": 7.916468158204576e-05,
      "loss": 1.1199,
      "step": 8044
    },
    {
      "epoch": 0.5799034095004685,
      "grad_norm": 0.2046489655103522,
      "learning_rate": 7.91418473836471e-05,
      "loss": 1.1028,
      "step": 8045
    },
    {
      "epoch": 0.5799754919628054,
      "grad_norm": 0.14096807022007865,
      "learning_rate": 7.911901432220435e-05,
      "loss": 1.071,
      "step": 8046
    },
    {
      "epoch": 0.5800475744251423,
      "grad_norm": 0.1808205399852678,
      "learning_rate": 7.909618239896216e-05,
      "loss": 0.9847,
      "step": 8047
    },
    {
      "epoch": 0.5801196568874792,
      "grad_norm": 0.11077916611628714,
      "learning_rate": 7.907335161516505e-05,
      "loss": 1.1186,
      "step": 8048
    },
    {
      "epoch": 0.5801917393498162,
      "grad_norm": 0.43812551609013645,
      "learning_rate": 7.905052197205747e-05,
      "loss": 1.0389,
      "step": 8049
    },
    {
      "epoch": 0.5802638218121531,
      "grad_norm": 0.16929940041906014,
      "learning_rate": 7.902769347088384e-05,
      "loss": 1.0496,
      "step": 8050
    },
    {
      "epoch": 0.58033590427449,
      "grad_norm": 0.11203869401744504,
      "learning_rate": 7.900486611288855e-05,
      "loss": 1.0732,
      "step": 8051
    },
    {
      "epoch": 0.5804079867368269,
      "grad_norm": 0.15318614871442043,
      "learning_rate": 7.898203989931592e-05,
      "loss": 1.0454,
      "step": 8052
    },
    {
      "epoch": 0.5804800691991638,
      "grad_norm": 0.11769554685637248,
      "learning_rate": 7.895921483141014e-05,
      "loss": 1.0138,
      "step": 8053
    },
    {
      "epoch": 0.5805521516615008,
      "grad_norm": 0.11419028404280882,
      "learning_rate": 7.893639091041536e-05,
      "loss": 1.026,
      "step": 8054
    },
    {
      "epoch": 0.5806242341238377,
      "grad_norm": 0.13859505160644303,
      "learning_rate": 7.89135681375757e-05,
      "loss": 1.027,
      "step": 8055
    },
    {
      "epoch": 0.5806963165861746,
      "grad_norm": 0.12318567602602015,
      "learning_rate": 7.889074651413527e-05,
      "loss": 1.029,
      "step": 8056
    },
    {
      "epoch": 0.5807683990485115,
      "grad_norm": 0.12147256713883385,
      "learning_rate": 7.886792604133799e-05,
      "loss": 1.0241,
      "step": 8057
    },
    {
      "epoch": 0.5808404815108484,
      "grad_norm": 0.13569180893691876,
      "learning_rate": 7.884510672042778e-05,
      "loss": 1.1307,
      "step": 8058
    },
    {
      "epoch": 0.5809125639731854,
      "grad_norm": 0.13138423821267523,
      "learning_rate": 7.882228855264849e-05,
      "loss": 1.075,
      "step": 8059
    },
    {
      "epoch": 0.5809846464355223,
      "grad_norm": 0.23691599070093033,
      "learning_rate": 7.879947153924399e-05,
      "loss": 1.0375,
      "step": 8060
    },
    {
      "epoch": 0.5810567288978592,
      "grad_norm": 0.16527039049350617,
      "learning_rate": 7.877665568145793e-05,
      "loss": 1.0823,
      "step": 8061
    },
    {
      "epoch": 0.5811288113601961,
      "grad_norm": 0.11322646923795196,
      "learning_rate": 7.8753840980534e-05,
      "loss": 0.9885,
      "step": 8062
    },
    {
      "epoch": 0.5812008938225329,
      "grad_norm": 0.13578797743812335,
      "learning_rate": 7.873102743771584e-05,
      "loss": 1.0445,
      "step": 8063
    },
    {
      "epoch": 0.5812729762848698,
      "grad_norm": 0.12755351176780075,
      "learning_rate": 7.87082150542469e-05,
      "loss": 1.0597,
      "step": 8064
    },
    {
      "epoch": 0.5813450587472068,
      "grad_norm": 0.11946979415904253,
      "learning_rate": 7.868540383137076e-05,
      "loss": 1.0542,
      "step": 8065
    },
    {
      "epoch": 0.5814171412095437,
      "grad_norm": 0.12163121730419207,
      "learning_rate": 7.866259377033081e-05,
      "loss": 1.0449,
      "step": 8066
    },
    {
      "epoch": 0.5814892236718806,
      "grad_norm": 0.12544256995541053,
      "learning_rate": 7.863978487237039e-05,
      "loss": 1.0816,
      "step": 8067
    },
    {
      "epoch": 0.5815613061342175,
      "grad_norm": 0.2720128116815735,
      "learning_rate": 7.861697713873274e-05,
      "loss": 1.0723,
      "step": 8068
    },
    {
      "epoch": 0.5816333885965544,
      "grad_norm": 0.11522629510445269,
      "learning_rate": 7.859417057066117e-05,
      "loss": 1.049,
      "step": 8069
    },
    {
      "epoch": 0.5817054710588914,
      "grad_norm": 0.1270208598520636,
      "learning_rate": 7.85713651693988e-05,
      "loss": 1.0461,
      "step": 8070
    },
    {
      "epoch": 0.5817775535212283,
      "grad_norm": 0.13214494999931084,
      "learning_rate": 7.854856093618877e-05,
      "loss": 1.013,
      "step": 8071
    },
    {
      "epoch": 0.5818496359835652,
      "grad_norm": 0.12688655679492464,
      "learning_rate": 7.852575787227401e-05,
      "loss": 1.0552,
      "step": 8072
    },
    {
      "epoch": 0.5819217184459021,
      "grad_norm": 0.12243248490529392,
      "learning_rate": 7.85029559788976e-05,
      "loss": 1.0868,
      "step": 8073
    },
    {
      "epoch": 0.581993800908239,
      "grad_norm": 0.11231630255525606,
      "learning_rate": 7.84801552573024e-05,
      "loss": 1.0637,
      "step": 8074
    },
    {
      "epoch": 0.582065883370576,
      "grad_norm": 0.13169066552251502,
      "learning_rate": 7.845735570873128e-05,
      "loss": 1.0083,
      "step": 8075
    },
    {
      "epoch": 0.5821379658329129,
      "grad_norm": 0.7155284677999951,
      "learning_rate": 7.843455733442698e-05,
      "loss": 1.0463,
      "step": 8076
    },
    {
      "epoch": 0.5822100482952498,
      "grad_norm": 0.12018946058382365,
      "learning_rate": 7.841176013563222e-05,
      "loss": 1.0467,
      "step": 8077
    },
    {
      "epoch": 0.5822821307575867,
      "grad_norm": 0.11174864400400353,
      "learning_rate": 7.838896411358969e-05,
      "loss": 1.125,
      "step": 8078
    },
    {
      "epoch": 0.5823542132199236,
      "grad_norm": 0.11990183244314898,
      "learning_rate": 7.836616926954196e-05,
      "loss": 1.0293,
      "step": 8079
    },
    {
      "epoch": 0.5824262956822605,
      "grad_norm": 0.13513791231099953,
      "learning_rate": 7.834337560473154e-05,
      "loss": 1.0203,
      "step": 8080
    },
    {
      "epoch": 0.5824983781445974,
      "grad_norm": 0.11520318532289071,
      "learning_rate": 7.832058312040087e-05,
      "loss": 0.9771,
      "step": 8081
    },
    {
      "epoch": 0.5825704606069343,
      "grad_norm": 0.12447166433602237,
      "learning_rate": 7.829779181779238e-05,
      "loss": 1.0216,
      "step": 8082
    },
    {
      "epoch": 0.5826425430692712,
      "grad_norm": 0.10548554101362449,
      "learning_rate": 7.827500169814842e-05,
      "loss": 1.0076,
      "step": 8083
    },
    {
      "epoch": 0.5827146255316081,
      "grad_norm": 0.1312172540441712,
      "learning_rate": 7.825221276271121e-05,
      "loss": 0.9966,
      "step": 8084
    },
    {
      "epoch": 0.582786707993945,
      "grad_norm": 0.10889071919184058,
      "learning_rate": 7.822942501272297e-05,
      "loss": 1.0566,
      "step": 8085
    },
    {
      "epoch": 0.582858790456282,
      "grad_norm": 0.11572366281828306,
      "learning_rate": 7.820663844942579e-05,
      "loss": 1.0448,
      "step": 8086
    },
    {
      "epoch": 0.5829308729186189,
      "grad_norm": 0.16494028445589484,
      "learning_rate": 7.818385307406184e-05,
      "loss": 0.94,
      "step": 8087
    },
    {
      "epoch": 0.5830029553809558,
      "grad_norm": 0.12480516425181666,
      "learning_rate": 7.816106888787304e-05,
      "loss": 0.9993,
      "step": 8088
    },
    {
      "epoch": 0.5830750378432927,
      "grad_norm": 0.1344784696532359,
      "learning_rate": 7.813828589210138e-05,
      "loss": 0.9446,
      "step": 8089
    },
    {
      "epoch": 0.5831471203056297,
      "grad_norm": 0.11688711066513534,
      "learning_rate": 7.811550408798867e-05,
      "loss": 0.9561,
      "step": 8090
    },
    {
      "epoch": 0.5832192027679666,
      "grad_norm": 0.11333371277742064,
      "learning_rate": 7.809272347677683e-05,
      "loss": 1.0562,
      "step": 8091
    },
    {
      "epoch": 0.5832912852303035,
      "grad_norm": 0.11911618644118588,
      "learning_rate": 7.806994405970753e-05,
      "loss": 1.0226,
      "step": 8092
    },
    {
      "epoch": 0.5833633676926404,
      "grad_norm": 0.11601660532913234,
      "learning_rate": 7.804716583802248e-05,
      "loss": 1.0629,
      "step": 8093
    },
    {
      "epoch": 0.5834354501549773,
      "grad_norm": 0.13420445898776134,
      "learning_rate": 7.802438881296323e-05,
      "loss": 0.9852,
      "step": 8094
    },
    {
      "epoch": 0.5835075326173143,
      "grad_norm": 0.1205297718508684,
      "learning_rate": 7.800161298577145e-05,
      "loss": 0.9887,
      "step": 8095
    },
    {
      "epoch": 0.5835796150796512,
      "grad_norm": 0.35663404596501186,
      "learning_rate": 7.797883835768856e-05,
      "loss": 1.071,
      "step": 8096
    },
    {
      "epoch": 0.583651697541988,
      "grad_norm": 0.11094872596128091,
      "learning_rate": 7.7956064929956e-05,
      "loss": 1.0159,
      "step": 8097
    },
    {
      "epoch": 0.5837237800043249,
      "grad_norm": 0.13402222456519167,
      "learning_rate": 7.793329270381508e-05,
      "loss": 1.0165,
      "step": 8098
    },
    {
      "epoch": 0.5837958624666618,
      "grad_norm": 0.23826060150354197,
      "learning_rate": 7.79105216805071e-05,
      "loss": 1.1046,
      "step": 8099
    },
    {
      "epoch": 0.5838679449289987,
      "grad_norm": 0.11274580561815851,
      "learning_rate": 7.788775186127335e-05,
      "loss": 0.9654,
      "step": 8100
    },
    {
      "epoch": 0.5839400273913357,
      "grad_norm": 0.18834801021201228,
      "learning_rate": 7.786498324735495e-05,
      "loss": 1.0839,
      "step": 8101
    },
    {
      "epoch": 0.5840121098536726,
      "grad_norm": 0.1090026793849704,
      "learning_rate": 7.784221583999298e-05,
      "loss": 1.0735,
      "step": 8102
    },
    {
      "epoch": 0.5840841923160095,
      "grad_norm": 0.12787122209534174,
      "learning_rate": 7.781944964042843e-05,
      "loss": 1.0015,
      "step": 8103
    },
    {
      "epoch": 0.5841562747783464,
      "grad_norm": 0.11964139054336054,
      "learning_rate": 7.779668464990235e-05,
      "loss": 1.0616,
      "step": 8104
    },
    {
      "epoch": 0.5842283572406833,
      "grad_norm": 0.129943490786469,
      "learning_rate": 7.77739208696556e-05,
      "loss": 1.0116,
      "step": 8105
    },
    {
      "epoch": 0.5843004397030203,
      "grad_norm": 0.11361270689342137,
      "learning_rate": 7.7751158300929e-05,
      "loss": 0.9515,
      "step": 8106
    },
    {
      "epoch": 0.5843725221653572,
      "grad_norm": 0.11867336238825016,
      "learning_rate": 7.772839694496332e-05,
      "loss": 1.0961,
      "step": 8107
    },
    {
      "epoch": 0.5844446046276941,
      "grad_norm": 0.17780850753655492,
      "learning_rate": 7.770563680299921e-05,
      "loss": 1.0369,
      "step": 8108
    },
    {
      "epoch": 0.584516687090031,
      "grad_norm": 0.11494926187890393,
      "learning_rate": 7.768287787627739e-05,
      "loss": 1.0964,
      "step": 8109
    },
    {
      "epoch": 0.5845887695523679,
      "grad_norm": 0.12467816109966957,
      "learning_rate": 7.766012016603836e-05,
      "loss": 1.0199,
      "step": 8110
    },
    {
      "epoch": 0.5846608520147049,
      "grad_norm": 0.12516500100459488,
      "learning_rate": 7.763736367352267e-05,
      "loss": 1.0637,
      "step": 8111
    },
    {
      "epoch": 0.5847329344770418,
      "grad_norm": 0.14039189607202832,
      "learning_rate": 7.761460839997065e-05,
      "loss": 1.1271,
      "step": 8112
    },
    {
      "epoch": 0.5848050169393787,
      "grad_norm": 0.3185237138048689,
      "learning_rate": 7.759185434662281e-05,
      "loss": 1.1101,
      "step": 8113
    },
    {
      "epoch": 0.5848770994017156,
      "grad_norm": 0.16733928166892273,
      "learning_rate": 7.756910151471935e-05,
      "loss": 0.9865,
      "step": 8114
    },
    {
      "epoch": 0.5849491818640524,
      "grad_norm": 0.13346232733741392,
      "learning_rate": 7.754634990550055e-05,
      "loss": 0.9729,
      "step": 8115
    },
    {
      "epoch": 0.5850212643263893,
      "grad_norm": 0.11563992167137521,
      "learning_rate": 7.752359952020649e-05,
      "loss": 1.0744,
      "step": 8116
    },
    {
      "epoch": 0.5850933467887263,
      "grad_norm": 0.11724108323834559,
      "learning_rate": 7.750085036007741e-05,
      "loss": 1.0441,
      "step": 8117
    },
    {
      "epoch": 0.5851654292510632,
      "grad_norm": 0.1907132584899814,
      "learning_rate": 7.747810242635325e-05,
      "loss": 0.9942,
      "step": 8118
    },
    {
      "epoch": 0.5852375117134001,
      "grad_norm": 0.1197451817544064,
      "learning_rate": 7.7455355720274e-05,
      "loss": 1.0434,
      "step": 8119
    },
    {
      "epoch": 0.585309594175737,
      "grad_norm": 0.13879670172789738,
      "learning_rate": 7.743261024307954e-05,
      "loss": 1.0766,
      "step": 8120
    },
    {
      "epoch": 0.5853816766380739,
      "grad_norm": 0.1265533664526477,
      "learning_rate": 7.740986599600969e-05,
      "loss": 1.0557,
      "step": 8121
    },
    {
      "epoch": 0.5854537591004109,
      "grad_norm": 0.11509240102077577,
      "learning_rate": 7.738712298030427e-05,
      "loss": 1.0075,
      "step": 8122
    },
    {
      "epoch": 0.5855258415627478,
      "grad_norm": 0.11932692795296557,
      "learning_rate": 7.736438119720297e-05,
      "loss": 1.06,
      "step": 8123
    },
    {
      "epoch": 0.5855979240250847,
      "grad_norm": 0.11711111247419526,
      "learning_rate": 7.734164064794537e-05,
      "loss": 0.9968,
      "step": 8124
    },
    {
      "epoch": 0.5856700064874216,
      "grad_norm": 0.12133705311764842,
      "learning_rate": 7.731890133377104e-05,
      "loss": 1.0598,
      "step": 8125
    },
    {
      "epoch": 0.5857420889497585,
      "grad_norm": 0.13236070950652068,
      "learning_rate": 7.729616325591951e-05,
      "loss": 1.1745,
      "step": 8126
    },
    {
      "epoch": 0.5858141714120955,
      "grad_norm": 0.14857020152453854,
      "learning_rate": 7.727342641563024e-05,
      "loss": 1.1179,
      "step": 8127
    },
    {
      "epoch": 0.5858862538744324,
      "grad_norm": 0.13259332113010808,
      "learning_rate": 7.725069081414252e-05,
      "loss": 1.0874,
      "step": 8128
    },
    {
      "epoch": 0.5859583363367693,
      "grad_norm": 0.1198435921473597,
      "learning_rate": 7.722795645269567e-05,
      "loss": 1.0043,
      "step": 8129
    },
    {
      "epoch": 0.5860304187991062,
      "grad_norm": 0.11664574682152708,
      "learning_rate": 7.720522333252889e-05,
      "loss": 1.0556,
      "step": 8130
    },
    {
      "epoch": 0.5861025012614431,
      "grad_norm": 0.1623272386622867,
      "learning_rate": 7.718249145488142e-05,
      "loss": 1.1543,
      "step": 8131
    },
    {
      "epoch": 0.58617458372378,
      "grad_norm": 0.21698738740672385,
      "learning_rate": 7.715976082099226e-05,
      "loss": 1.037,
      "step": 8132
    },
    {
      "epoch": 0.5862466661861169,
      "grad_norm": 0.10422726940627078,
      "learning_rate": 7.71370314321005e-05,
      "loss": 1.0715,
      "step": 8133
    },
    {
      "epoch": 0.5863187486484538,
      "grad_norm": 0.20610690642511442,
      "learning_rate": 7.711430328944501e-05,
      "loss": 1.034,
      "step": 8134
    },
    {
      "epoch": 0.5863908311107907,
      "grad_norm": 0.12683510726881192,
      "learning_rate": 7.709157639426478e-05,
      "loss": 1.0409,
      "step": 8135
    },
    {
      "epoch": 0.5864629135731276,
      "grad_norm": 0.10818636300831053,
      "learning_rate": 7.706885074779859e-05,
      "loss": 1.0106,
      "step": 8136
    },
    {
      "epoch": 0.5865349960354646,
      "grad_norm": 0.23711491742684299,
      "learning_rate": 7.704612635128515e-05,
      "loss": 1.0506,
      "step": 8137
    },
    {
      "epoch": 0.5866070784978015,
      "grad_norm": 0.15513655002068685,
      "learning_rate": 7.702340320596316e-05,
      "loss": 1.0381,
      "step": 8138
    },
    {
      "epoch": 0.5866791609601384,
      "grad_norm": 0.13312162206941916,
      "learning_rate": 7.70006813130713e-05,
      "loss": 1.0506,
      "step": 8139
    },
    {
      "epoch": 0.5867512434224753,
      "grad_norm": 0.12832746582358598,
      "learning_rate": 7.697796067384804e-05,
      "loss": 1.0544,
      "step": 8140
    },
    {
      "epoch": 0.5868233258848122,
      "grad_norm": 0.14104664166621675,
      "learning_rate": 7.695524128953189e-05,
      "loss": 0.9618,
      "step": 8141
    },
    {
      "epoch": 0.5868954083471492,
      "grad_norm": 0.14960253124678244,
      "learning_rate": 7.693252316136127e-05,
      "loss": 1.0119,
      "step": 8142
    },
    {
      "epoch": 0.5869674908094861,
      "grad_norm": 0.16058822160192274,
      "learning_rate": 7.690980629057447e-05,
      "loss": 1.1407,
      "step": 8143
    },
    {
      "epoch": 0.587039573271823,
      "grad_norm": 0.11680274236924902,
      "learning_rate": 7.68870906784098e-05,
      "loss": 0.9908,
      "step": 8144
    },
    {
      "epoch": 0.5871116557341599,
      "grad_norm": 0.10778513664409821,
      "learning_rate": 7.686437632610548e-05,
      "loss": 1.0699,
      "step": 8145
    },
    {
      "epoch": 0.5871837381964968,
      "grad_norm": 0.5237337956690794,
      "learning_rate": 7.684166323489965e-05,
      "loss": 1.0976,
      "step": 8146
    },
    {
      "epoch": 0.5872558206588337,
      "grad_norm": 0.1240704614110086,
      "learning_rate": 7.68189514060303e-05,
      "loss": 1.0875,
      "step": 8147
    },
    {
      "epoch": 0.5873279031211707,
      "grad_norm": 0.12688216483388517,
      "learning_rate": 7.679624084073551e-05,
      "loss": 1.0265,
      "step": 8148
    },
    {
      "epoch": 0.5873999855835075,
      "grad_norm": 0.10675597918272065,
      "learning_rate": 7.677353154025317e-05,
      "loss": 1.0,
      "step": 8149
    },
    {
      "epoch": 0.5874720680458444,
      "grad_norm": 0.11802463583821363,
      "learning_rate": 7.675082350582119e-05,
      "loss": 1.0236,
      "step": 8150
    },
    {
      "epoch": 0.5875441505081813,
      "grad_norm": 0.11571293834057984,
      "learning_rate": 7.672811673867729e-05,
      "loss": 1.025,
      "step": 8151
    },
    {
      "epoch": 0.5876162329705182,
      "grad_norm": 0.144480391402815,
      "learning_rate": 7.670541124005921e-05,
      "loss": 1.1083,
      "step": 8152
    },
    {
      "epoch": 0.5876883154328552,
      "grad_norm": 0.1675072328004864,
      "learning_rate": 7.668270701120464e-05,
      "loss": 1.0522,
      "step": 8153
    },
    {
      "epoch": 0.5877603978951921,
      "grad_norm": 0.10989291819165631,
      "learning_rate": 7.666000405335115e-05,
      "loss": 1.0913,
      "step": 8154
    },
    {
      "epoch": 0.587832480357529,
      "grad_norm": 0.12699627895106852,
      "learning_rate": 7.663730236773623e-05,
      "loss": 1.0413,
      "step": 8155
    },
    {
      "epoch": 0.5879045628198659,
      "grad_norm": 0.11339613181888102,
      "learning_rate": 7.661460195559732e-05,
      "loss": 0.9872,
      "step": 8156
    },
    {
      "epoch": 0.5879766452822028,
      "grad_norm": 0.17553673907476777,
      "learning_rate": 7.659190281817182e-05,
      "loss": 1.0374,
      "step": 8157
    },
    {
      "epoch": 0.5880487277445398,
      "grad_norm": 0.15069203541834522,
      "learning_rate": 7.656920495669707e-05,
      "loss": 1.009,
      "step": 8158
    },
    {
      "epoch": 0.5881208102068767,
      "grad_norm": 0.16044448359738273,
      "learning_rate": 7.654650837241024e-05,
      "loss": 1.0298,
      "step": 8159
    },
    {
      "epoch": 0.5881928926692136,
      "grad_norm": 0.12906468259772852,
      "learning_rate": 7.652381306654851e-05,
      "loss": 1.0392,
      "step": 8160
    },
    {
      "epoch": 0.5882649751315505,
      "grad_norm": 0.23868903524229101,
      "learning_rate": 7.6501119040349e-05,
      "loss": 1.0934,
      "step": 8161
    },
    {
      "epoch": 0.5883370575938874,
      "grad_norm": 0.1368368415798839,
      "learning_rate": 7.647842629504876e-05,
      "loss": 1.0806,
      "step": 8162
    },
    {
      "epoch": 0.5884091400562244,
      "grad_norm": 0.12598098368693073,
      "learning_rate": 7.645573483188471e-05,
      "loss": 1.0339,
      "step": 8163
    },
    {
      "epoch": 0.5884812225185613,
      "grad_norm": 0.12601678880066106,
      "learning_rate": 7.643304465209374e-05,
      "loss": 1.0104,
      "step": 8164
    },
    {
      "epoch": 0.5885533049808982,
      "grad_norm": 0.11089752070924451,
      "learning_rate": 7.641035575691263e-05,
      "loss": 0.9404,
      "step": 8165
    },
    {
      "epoch": 0.5886253874432351,
      "grad_norm": 0.12041353256844507,
      "learning_rate": 7.638766814757821e-05,
      "loss": 1.0321,
      "step": 8166
    },
    {
      "epoch": 0.5886974699055719,
      "grad_norm": 0.20864669887871704,
      "learning_rate": 7.636498182532712e-05,
      "loss": 0.9874,
      "step": 8167
    },
    {
      "epoch": 0.5887695523679088,
      "grad_norm": 0.13657057619458332,
      "learning_rate": 7.634229679139598e-05,
      "loss": 1.0126,
      "step": 8168
    },
    {
      "epoch": 0.5888416348302458,
      "grad_norm": 0.11141411415448714,
      "learning_rate": 7.631961304702124e-05,
      "loss": 1.0408,
      "step": 8169
    },
    {
      "epoch": 0.5889137172925827,
      "grad_norm": 0.12729434725791922,
      "learning_rate": 7.629693059343951e-05,
      "loss": 1.0746,
      "step": 8170
    },
    {
      "epoch": 0.5889857997549196,
      "grad_norm": 0.4250085272406547,
      "learning_rate": 7.627424943188709e-05,
      "loss": 1.1646,
      "step": 8171
    },
    {
      "epoch": 0.5890578822172565,
      "grad_norm": 0.11508982810778716,
      "learning_rate": 7.625156956360036e-05,
      "loss": 0.9559,
      "step": 8172
    },
    {
      "epoch": 0.5891299646795934,
      "grad_norm": 0.11852192175089132,
      "learning_rate": 7.622889098981553e-05,
      "loss": 1.0692,
      "step": 8173
    },
    {
      "epoch": 0.5892020471419304,
      "grad_norm": 0.11760377132168302,
      "learning_rate": 7.620621371176878e-05,
      "loss": 1.0218,
      "step": 8174
    },
    {
      "epoch": 0.5892741296042673,
      "grad_norm": 0.10635614384216249,
      "learning_rate": 7.618353773069626e-05,
      "loss": 1.0351,
      "step": 8175
    },
    {
      "epoch": 0.5893462120666042,
      "grad_norm": 0.1751237656670311,
      "learning_rate": 7.616086304783403e-05,
      "loss": 1.0545,
      "step": 8176
    },
    {
      "epoch": 0.5894182945289411,
      "grad_norm": 0.3065866258487919,
      "learning_rate": 7.613818966441801e-05,
      "loss": 1.069,
      "step": 8177
    },
    {
      "epoch": 0.589490376991278,
      "grad_norm": 0.12580246366516504,
      "learning_rate": 7.611551758168412e-05,
      "loss": 1.0751,
      "step": 8178
    },
    {
      "epoch": 0.589562459453615,
      "grad_norm": 0.1186961547685137,
      "learning_rate": 7.60928468008682e-05,
      "loss": 1.0128,
      "step": 8179
    },
    {
      "epoch": 0.5896345419159519,
      "grad_norm": 0.18994289785392648,
      "learning_rate": 7.607017732320604e-05,
      "loss": 1.0327,
      "step": 8180
    },
    {
      "epoch": 0.5897066243782888,
      "grad_norm": 0.1402343571418674,
      "learning_rate": 7.604750914993329e-05,
      "loss": 1.0231,
      "step": 8181
    },
    {
      "epoch": 0.5897787068406257,
      "grad_norm": 0.11892886282659089,
      "learning_rate": 7.602484228228552e-05,
      "loss": 0.9899,
      "step": 8182
    },
    {
      "epoch": 0.5898507893029626,
      "grad_norm": 0.4209927545770812,
      "learning_rate": 7.600217672149839e-05,
      "loss": 1.1104,
      "step": 8183
    },
    {
      "epoch": 0.5899228717652995,
      "grad_norm": 0.14887850305348682,
      "learning_rate": 7.597951246880732e-05,
      "loss": 0.9765,
      "step": 8184
    },
    {
      "epoch": 0.5899949542276364,
      "grad_norm": 0.18298014501691143,
      "learning_rate": 7.59568495254477e-05,
      "loss": 1.0517,
      "step": 8185
    },
    {
      "epoch": 0.5900670366899733,
      "grad_norm": 0.1868918294397054,
      "learning_rate": 7.59341878926549e-05,
      "loss": 1.088,
      "step": 8186
    },
    {
      "epoch": 0.5901391191523102,
      "grad_norm": 0.16788193282171318,
      "learning_rate": 7.591152757166409e-05,
      "loss": 1.011,
      "step": 8187
    },
    {
      "epoch": 0.5902112016146471,
      "grad_norm": 0.12761572511624317,
      "learning_rate": 7.58888685637106e-05,
      "loss": 0.9992,
      "step": 8188
    },
    {
      "epoch": 0.590283284076984,
      "grad_norm": 1.5228630162137509,
      "learning_rate": 7.586621087002945e-05,
      "loss": 1.0793,
      "step": 8189
    },
    {
      "epoch": 0.590355366539321,
      "grad_norm": 0.1415177409454454,
      "learning_rate": 7.584355449185574e-05,
      "loss": 1.0476,
      "step": 8190
    },
    {
      "epoch": 0.5904274490016579,
      "grad_norm": 0.284990037778724,
      "learning_rate": 7.582089943042435e-05,
      "loss": 0.9903,
      "step": 8191
    },
    {
      "epoch": 0.5904995314639948,
      "grad_norm": 0.1208010827470453,
      "learning_rate": 7.579824568697032e-05,
      "loss": 1.023,
      "step": 8192
    },
    {
      "epoch": 0.5905716139263317,
      "grad_norm": 0.11882526338908837,
      "learning_rate": 7.57755932627284e-05,
      "loss": 1.0992,
      "step": 8193
    },
    {
      "epoch": 0.5906436963886686,
      "grad_norm": 0.15169844813736508,
      "learning_rate": 7.575294215893337e-05,
      "loss": 0.9949,
      "step": 8194
    },
    {
      "epoch": 0.5907157788510056,
      "grad_norm": 0.2356773003102037,
      "learning_rate": 7.573029237681991e-05,
      "loss": 1.0345,
      "step": 8195
    },
    {
      "epoch": 0.5907878613133425,
      "grad_norm": 0.11080251470964171,
      "learning_rate": 7.57076439176226e-05,
      "loss": 0.9901,
      "step": 8196
    },
    {
      "epoch": 0.5908599437756794,
      "grad_norm": 0.12539922090400668,
      "learning_rate": 7.568499678257605e-05,
      "loss": 1.0618,
      "step": 8197
    },
    {
      "epoch": 0.5909320262380163,
      "grad_norm": 0.1511006945132873,
      "learning_rate": 7.566235097291475e-05,
      "loss": 1.0233,
      "step": 8198
    },
    {
      "epoch": 0.5910041087003532,
      "grad_norm": 0.15792697691621269,
      "learning_rate": 7.563970648987299e-05,
      "loss": 0.9928,
      "step": 8199
    },
    {
      "epoch": 0.5910761911626902,
      "grad_norm": 0.11104492410187893,
      "learning_rate": 7.561706333468517e-05,
      "loss": 1.0388,
      "step": 8200
    },
    {
      "epoch": 0.591148273625027,
      "grad_norm": 0.11159545869386221,
      "learning_rate": 7.559442150858553e-05,
      "loss": 1.0513,
      "step": 8201
    },
    {
      "epoch": 0.5912203560873639,
      "grad_norm": 0.1145249982301263,
      "learning_rate": 7.55717810128083e-05,
      "loss": 1.0963,
      "step": 8202
    },
    {
      "epoch": 0.5912924385497008,
      "grad_norm": 0.12440737129637607,
      "learning_rate": 7.554914184858751e-05,
      "loss": 1.0246,
      "step": 8203
    },
    {
      "epoch": 0.5913645210120377,
      "grad_norm": 0.14425817041254088,
      "learning_rate": 7.552650401715722e-05,
      "loss": 1.0203,
      "step": 8204
    },
    {
      "epoch": 0.5914366034743747,
      "grad_norm": 0.13640172152990804,
      "learning_rate": 7.550386751975143e-05,
      "loss": 1.0976,
      "step": 8205
    },
    {
      "epoch": 0.5915086859367116,
      "grad_norm": 0.18657342019891984,
      "learning_rate": 7.548123235760403e-05,
      "loss": 1.0714,
      "step": 8206
    },
    {
      "epoch": 0.5915807683990485,
      "grad_norm": 0.1426090925278554,
      "learning_rate": 7.545859853194878e-05,
      "loss": 1.0753,
      "step": 8207
    },
    {
      "epoch": 0.5916528508613854,
      "grad_norm": 0.6890846148305917,
      "learning_rate": 7.543596604401948e-05,
      "loss": 1.0458,
      "step": 8208
    },
    {
      "epoch": 0.5917249333237223,
      "grad_norm": 0.11470224297332576,
      "learning_rate": 7.541333489504975e-05,
      "loss": 0.9935,
      "step": 8209
    },
    {
      "epoch": 0.5917970157860593,
      "grad_norm": 0.12899066014763683,
      "learning_rate": 7.53907050862733e-05,
      "loss": 0.9975,
      "step": 8210
    },
    {
      "epoch": 0.5918690982483962,
      "grad_norm": 0.11227920564957529,
      "learning_rate": 7.536807661892353e-05,
      "loss": 1.0102,
      "step": 8211
    },
    {
      "epoch": 0.5919411807107331,
      "grad_norm": 0.15824230009427595,
      "learning_rate": 7.534544949423397e-05,
      "loss": 1.0787,
      "step": 8212
    },
    {
      "epoch": 0.59201326317307,
      "grad_norm": 0.4850759412847989,
      "learning_rate": 7.532282371343795e-05,
      "loss": 1.0984,
      "step": 8213
    },
    {
      "epoch": 0.5920853456354069,
      "grad_norm": 0.9994939035612062,
      "learning_rate": 7.530019927776885e-05,
      "loss": 1.02,
      "step": 8214
    },
    {
      "epoch": 0.5921574280977439,
      "grad_norm": 0.108555464949087,
      "learning_rate": 7.527757618845984e-05,
      "loss": 0.979,
      "step": 8215
    },
    {
      "epoch": 0.5922295105600808,
      "grad_norm": 0.11208334265043983,
      "learning_rate": 7.525495444674409e-05,
      "loss": 1.041,
      "step": 8216
    },
    {
      "epoch": 0.5923015930224177,
      "grad_norm": 0.1399789271640789,
      "learning_rate": 7.523233405385473e-05,
      "loss": 0.9907,
      "step": 8217
    },
    {
      "epoch": 0.5923736754847546,
      "grad_norm": 0.16369125774704366,
      "learning_rate": 7.520971501102469e-05,
      "loss": 1.0087,
      "step": 8218
    },
    {
      "epoch": 0.5924457579470914,
      "grad_norm": 0.16224298188620875,
      "learning_rate": 7.518709731948699e-05,
      "loss": 1.0219,
      "step": 8219
    },
    {
      "epoch": 0.5925178404094283,
      "grad_norm": 0.18340921766127682,
      "learning_rate": 7.516448098047446e-05,
      "loss": 1.042,
      "step": 8220
    },
    {
      "epoch": 0.5925899228717653,
      "grad_norm": 0.12287476519072717,
      "learning_rate": 7.514186599521993e-05,
      "loss": 0.9934,
      "step": 8221
    },
    {
      "epoch": 0.5926620053341022,
      "grad_norm": 0.14920385974677133,
      "learning_rate": 7.511925236495605e-05,
      "loss": 1.0243,
      "step": 8222
    },
    {
      "epoch": 0.5927340877964391,
      "grad_norm": 0.25280767020796974,
      "learning_rate": 7.509664009091552e-05,
      "loss": 1.0608,
      "step": 8223
    },
    {
      "epoch": 0.592806170258776,
      "grad_norm": 0.19083313612583563,
      "learning_rate": 7.507402917433091e-05,
      "loss": 1.0877,
      "step": 8224
    },
    {
      "epoch": 0.5928782527211129,
      "grad_norm": 0.1329639807205813,
      "learning_rate": 7.505141961643471e-05,
      "loss": 1.0495,
      "step": 8225
    },
    {
      "epoch": 0.5929503351834499,
      "grad_norm": 0.10431152546109296,
      "learning_rate": 7.502881141845929e-05,
      "loss": 1.0851,
      "step": 8226
    },
    {
      "epoch": 0.5930224176457868,
      "grad_norm": 0.12779169882404068,
      "learning_rate": 7.500620458163709e-05,
      "loss": 1.0696,
      "step": 8227
    },
    {
      "epoch": 0.5930945001081237,
      "grad_norm": 0.11421999060121384,
      "learning_rate": 7.498359910720034e-05,
      "loss": 1.067,
      "step": 8228
    },
    {
      "epoch": 0.5931665825704606,
      "grad_norm": 0.12849119483177193,
      "learning_rate": 7.496099499638125e-05,
      "loss": 1.046,
      "step": 8229
    },
    {
      "epoch": 0.5932386650327975,
      "grad_norm": 0.1382058909473472,
      "learning_rate": 7.493839225041193e-05,
      "loss": 0.9614,
      "step": 8230
    },
    {
      "epoch": 0.5933107474951345,
      "grad_norm": 0.11777331538180741,
      "learning_rate": 7.491579087052441e-05,
      "loss": 1.0244,
      "step": 8231
    },
    {
      "epoch": 0.5933828299574714,
      "grad_norm": 0.1331961609885304,
      "learning_rate": 7.48931908579507e-05,
      "loss": 1.0626,
      "step": 8232
    },
    {
      "epoch": 0.5934549124198083,
      "grad_norm": 0.13780079894072134,
      "learning_rate": 7.487059221392274e-05,
      "loss": 1.0075,
      "step": 8233
    },
    {
      "epoch": 0.5935269948821452,
      "grad_norm": 0.11478718118026728,
      "learning_rate": 7.48479949396723e-05,
      "loss": 1.0765,
      "step": 8234
    },
    {
      "epoch": 0.5935990773444821,
      "grad_norm": 0.16578314704852934,
      "learning_rate": 7.482539903643114e-05,
      "loss": 1.0927,
      "step": 8235
    },
    {
      "epoch": 0.593671159806819,
      "grad_norm": 0.12259877018653247,
      "learning_rate": 7.480280450543095e-05,
      "loss": 1.0549,
      "step": 8236
    },
    {
      "epoch": 0.5937432422691559,
      "grad_norm": 0.13510704378053062,
      "learning_rate": 7.478021134790336e-05,
      "loss": 1.0545,
      "step": 8237
    },
    {
      "epoch": 0.5938153247314928,
      "grad_norm": 0.11277441501774434,
      "learning_rate": 7.475761956507987e-05,
      "loss": 0.9904,
      "step": 8238
    },
    {
      "epoch": 0.5938874071938297,
      "grad_norm": 0.10331731776545072,
      "learning_rate": 7.473502915819194e-05,
      "loss": 1.0275,
      "step": 8239
    },
    {
      "epoch": 0.5939594896561666,
      "grad_norm": 0.12324985425054363,
      "learning_rate": 7.47124401284709e-05,
      "loss": 0.9842,
      "step": 8240
    },
    {
      "epoch": 0.5940315721185035,
      "grad_norm": 0.10830624898324351,
      "learning_rate": 7.468985247714815e-05,
      "loss": 1.0231,
      "step": 8241
    },
    {
      "epoch": 0.5941036545808405,
      "grad_norm": 0.12503586522308827,
      "learning_rate": 7.466726620545485e-05,
      "loss": 1.073,
      "step": 8242
    },
    {
      "epoch": 0.5941757370431774,
      "grad_norm": 0.13204316174373873,
      "learning_rate": 7.464468131462222e-05,
      "loss": 0.985,
      "step": 8243
    },
    {
      "epoch": 0.5942478195055143,
      "grad_norm": 0.1099362906628396,
      "learning_rate": 7.462209780588121e-05,
      "loss": 1.0363,
      "step": 8244
    },
    {
      "epoch": 0.5943199019678512,
      "grad_norm": 0.11126222434845086,
      "learning_rate": 7.459951568046297e-05,
      "loss": 0.9954,
      "step": 8245
    },
    {
      "epoch": 0.5943919844301881,
      "grad_norm": 0.22593863355881605,
      "learning_rate": 7.457693493959834e-05,
      "loss": 0.9943,
      "step": 8246
    },
    {
      "epoch": 0.5944640668925251,
      "grad_norm": 0.13583621511522256,
      "learning_rate": 7.455435558451823e-05,
      "loss": 1.0074,
      "step": 8247
    },
    {
      "epoch": 0.594536149354862,
      "grad_norm": 0.12286778915862606,
      "learning_rate": 7.453177761645331e-05,
      "loss": 1.0535,
      "step": 8248
    },
    {
      "epoch": 0.5946082318171989,
      "grad_norm": 0.1161552936582278,
      "learning_rate": 7.450920103663443e-05,
      "loss": 1.0201,
      "step": 8249
    },
    {
      "epoch": 0.5946803142795358,
      "grad_norm": 0.10235299905983704,
      "learning_rate": 7.448662584629211e-05,
      "loss": 1.0416,
      "step": 8250
    },
    {
      "epoch": 0.5947523967418727,
      "grad_norm": 0.11211018408122186,
      "learning_rate": 7.446405204665697e-05,
      "loss": 1.0472,
      "step": 8251
    },
    {
      "epoch": 0.5948244792042097,
      "grad_norm": 0.13208433855500135,
      "learning_rate": 7.44414796389594e-05,
      "loss": 1.0939,
      "step": 8252
    },
    {
      "epoch": 0.5948965616665465,
      "grad_norm": 0.14559739380278772,
      "learning_rate": 7.441890862442984e-05,
      "loss": 1.0222,
      "step": 8253
    },
    {
      "epoch": 0.5949686441288834,
      "grad_norm": 0.10475012493862379,
      "learning_rate": 7.439633900429862e-05,
      "loss": 1.0673,
      "step": 8254
    },
    {
      "epoch": 0.5950407265912203,
      "grad_norm": 0.11697771928162073,
      "learning_rate": 7.437377077979601e-05,
      "loss": 1.019,
      "step": 8255
    },
    {
      "epoch": 0.5951128090535572,
      "grad_norm": 0.16480943042281448,
      "learning_rate": 7.435120395215214e-05,
      "loss": 1.0397,
      "step": 8256
    },
    {
      "epoch": 0.5951848915158942,
      "grad_norm": 0.12498247739325836,
      "learning_rate": 7.432863852259708e-05,
      "loss": 1.1011,
      "step": 8257
    },
    {
      "epoch": 0.5952569739782311,
      "grad_norm": 0.16566842602237372,
      "learning_rate": 7.43060744923609e-05,
      "loss": 0.9983,
      "step": 8258
    },
    {
      "epoch": 0.595329056440568,
      "grad_norm": 0.13847071971972644,
      "learning_rate": 7.428351186267356e-05,
      "loss": 1.063,
      "step": 8259
    },
    {
      "epoch": 0.5954011389029049,
      "grad_norm": 0.11923163898458199,
      "learning_rate": 7.426095063476486e-05,
      "loss": 1.0346,
      "step": 8260
    },
    {
      "epoch": 0.5954732213652418,
      "grad_norm": 0.12437099526096204,
      "learning_rate": 7.423839080986464e-05,
      "loss": 1.0427,
      "step": 8261
    },
    {
      "epoch": 0.5955453038275788,
      "grad_norm": 0.1240077978996466,
      "learning_rate": 7.421583238920253e-05,
      "loss": 1.0675,
      "step": 8262
    },
    {
      "epoch": 0.5956173862899157,
      "grad_norm": 0.1265620657396026,
      "learning_rate": 7.419327537400829e-05,
      "loss": 1.0137,
      "step": 8263
    },
    {
      "epoch": 0.5956894687522526,
      "grad_norm": 0.12044277667051599,
      "learning_rate": 7.41707197655114e-05,
      "loss": 0.9645,
      "step": 8264
    },
    {
      "epoch": 0.5957615512145895,
      "grad_norm": 0.12058425199929652,
      "learning_rate": 7.414816556494138e-05,
      "loss": 0.996,
      "step": 8265
    },
    {
      "epoch": 0.5958336336769264,
      "grad_norm": 0.13201742343722103,
      "learning_rate": 7.412561277352753e-05,
      "loss": 0.96,
      "step": 8266
    },
    {
      "epoch": 0.5959057161392634,
      "grad_norm": 0.23010792392498788,
      "learning_rate": 7.410306139249934e-05,
      "loss": 1.0016,
      "step": 8267
    },
    {
      "epoch": 0.5959777986016003,
      "grad_norm": 0.13089781380478502,
      "learning_rate": 7.408051142308596e-05,
      "loss": 0.9918,
      "step": 8268
    },
    {
      "epoch": 0.5960498810639372,
      "grad_norm": 0.11906649283447637,
      "learning_rate": 7.405796286651662e-05,
      "loss": 1.0407,
      "step": 8269
    },
    {
      "epoch": 0.5961219635262741,
      "grad_norm": 0.1267262160532673,
      "learning_rate": 7.403541572402036e-05,
      "loss": 0.9852,
      "step": 8270
    },
    {
      "epoch": 0.5961940459886109,
      "grad_norm": 0.12661052756887295,
      "learning_rate": 7.40128699968262e-05,
      "loss": 1.0599,
      "step": 8271
    },
    {
      "epoch": 0.5962661284509478,
      "grad_norm": 0.11421537003592559,
      "learning_rate": 7.399032568616315e-05,
      "loss": 1.0834,
      "step": 8272
    },
    {
      "epoch": 0.5963382109132848,
      "grad_norm": 0.11832014595682254,
      "learning_rate": 7.396778279326005e-05,
      "loss": 1.0347,
      "step": 8273
    },
    {
      "epoch": 0.5964102933756217,
      "grad_norm": 0.1148010293052031,
      "learning_rate": 7.394524131934566e-05,
      "loss": 1.0236,
      "step": 8274
    },
    {
      "epoch": 0.5964823758379586,
      "grad_norm": 0.12791840594809878,
      "learning_rate": 7.392270126564869e-05,
      "loss": 1.0513,
      "step": 8275
    },
    {
      "epoch": 0.5965544583002955,
      "grad_norm": 0.12439120276138416,
      "learning_rate": 7.390016263339782e-05,
      "loss": 1.0668,
      "step": 8276
    },
    {
      "epoch": 0.5966265407626324,
      "grad_norm": 1.2464857329080183,
      "learning_rate": 7.387762542382159e-05,
      "loss": 1.0861,
      "step": 8277
    },
    {
      "epoch": 0.5966986232249694,
      "grad_norm": 0.12181006697871725,
      "learning_rate": 7.385508963814847e-05,
      "loss": 1.0411,
      "step": 8278
    },
    {
      "epoch": 0.5967707056873063,
      "grad_norm": 0.11484383565854726,
      "learning_rate": 7.383255527760683e-05,
      "loss": 1.0339,
      "step": 8279
    },
    {
      "epoch": 0.5968427881496432,
      "grad_norm": 0.12078992223164636,
      "learning_rate": 7.381002234342506e-05,
      "loss": 1.0538,
      "step": 8280
    },
    {
      "epoch": 0.5969148706119801,
      "grad_norm": 0.11401868982438254,
      "learning_rate": 7.378749083683139e-05,
      "loss": 1.0674,
      "step": 8281
    },
    {
      "epoch": 0.596986953074317,
      "grad_norm": 0.22068559739017568,
      "learning_rate": 7.376496075905395e-05,
      "loss": 1.043,
      "step": 8282
    },
    {
      "epoch": 0.597059035536654,
      "grad_norm": 0.11015157019862325,
      "learning_rate": 7.374243211132085e-05,
      "loss": 0.9555,
      "step": 8283
    },
    {
      "epoch": 0.5971311179989909,
      "grad_norm": 0.11627235205864828,
      "learning_rate": 7.37199048948601e-05,
      "loss": 1.0726,
      "step": 8284
    },
    {
      "epoch": 0.5972032004613278,
      "grad_norm": 0.13150338326250177,
      "learning_rate": 7.369737911089967e-05,
      "loss": 1.0286,
      "step": 8285
    },
    {
      "epoch": 0.5972752829236647,
      "grad_norm": 0.26729694624327843,
      "learning_rate": 7.367485476066736e-05,
      "loss": 1.0341,
      "step": 8286
    },
    {
      "epoch": 0.5973473653860016,
      "grad_norm": 0.10710638838106186,
      "learning_rate": 7.3652331845391e-05,
      "loss": 0.9894,
      "step": 8287
    },
    {
      "epoch": 0.5974194478483384,
      "grad_norm": 0.15964572059585969,
      "learning_rate": 7.362981036629823e-05,
      "loss": 1.0234,
      "step": 8288
    },
    {
      "epoch": 0.5974915303106754,
      "grad_norm": 0.18722980671866662,
      "learning_rate": 7.360729032461675e-05,
      "loss": 1.1021,
      "step": 8289
    },
    {
      "epoch": 0.5975636127730123,
      "grad_norm": 0.11625952906283407,
      "learning_rate": 7.358477172157405e-05,
      "loss": 1.0402,
      "step": 8290
    },
    {
      "epoch": 0.5976356952353492,
      "grad_norm": 0.15597610703565107,
      "learning_rate": 7.35622545583976e-05,
      "loss": 1.1212,
      "step": 8291
    },
    {
      "epoch": 0.5977077776976861,
      "grad_norm": 0.10583061850673162,
      "learning_rate": 7.353973883631483e-05,
      "loss": 1.0139,
      "step": 8292
    },
    {
      "epoch": 0.597779860160023,
      "grad_norm": 0.12006624988544035,
      "learning_rate": 7.351722455655297e-05,
      "loss": 1.0366,
      "step": 8293
    },
    {
      "epoch": 0.59785194262236,
      "grad_norm": 0.12257095423909536,
      "learning_rate": 7.34947117203393e-05,
      "loss": 1.0287,
      "step": 8294
    },
    {
      "epoch": 0.5979240250846969,
      "grad_norm": 0.26314426998229795,
      "learning_rate": 7.347220032890097e-05,
      "loss": 1.0301,
      "step": 8295
    },
    {
      "epoch": 0.5979961075470338,
      "grad_norm": 0.10961806072877747,
      "learning_rate": 7.344969038346507e-05,
      "loss": 1.0177,
      "step": 8296
    },
    {
      "epoch": 0.5980681900093707,
      "grad_norm": 0.39952136595951754,
      "learning_rate": 7.342718188525854e-05,
      "loss": 1.0668,
      "step": 8297
    },
    {
      "epoch": 0.5981402724717076,
      "grad_norm": 0.11288377384263112,
      "learning_rate": 7.340467483550832e-05,
      "loss": 1.0601,
      "step": 8298
    },
    {
      "epoch": 0.5982123549340446,
      "grad_norm": 0.1139087713148862,
      "learning_rate": 7.338216923544128e-05,
      "loss": 1.0913,
      "step": 8299
    },
    {
      "epoch": 0.5982844373963815,
      "grad_norm": 0.12569342196313818,
      "learning_rate": 7.335966508628414e-05,
      "loss": 1.0298,
      "step": 8300
    },
    {
      "epoch": 0.5983565198587184,
      "grad_norm": 0.13684976391807846,
      "learning_rate": 7.333716238926355e-05,
      "loss": 1.0186,
      "step": 8301
    },
    {
      "epoch": 0.5984286023210553,
      "grad_norm": 0.1355454989270519,
      "learning_rate": 7.331466114560618e-05,
      "loss": 1.0605,
      "step": 8302
    },
    {
      "epoch": 0.5985006847833922,
      "grad_norm": 0.1890594928532361,
      "learning_rate": 7.329216135653848e-05,
      "loss": 1.1727,
      "step": 8303
    },
    {
      "epoch": 0.5985727672457292,
      "grad_norm": 0.18232328925986124,
      "learning_rate": 7.326966302328697e-05,
      "loss": 1.0078,
      "step": 8304
    },
    {
      "epoch": 0.598644849708066,
      "grad_norm": 0.14662400635591422,
      "learning_rate": 7.324716614707793e-05,
      "loss": 0.983,
      "step": 8305
    },
    {
      "epoch": 0.5987169321704029,
      "grad_norm": 0.13451007392215378,
      "learning_rate": 7.322467072913765e-05,
      "loss": 0.9943,
      "step": 8306
    },
    {
      "epoch": 0.5987890146327398,
      "grad_norm": 0.11639312410663734,
      "learning_rate": 7.320217677069238e-05,
      "loss": 1.0859,
      "step": 8307
    },
    {
      "epoch": 0.5988610970950767,
      "grad_norm": 0.12988079707894684,
      "learning_rate": 7.317968427296824e-05,
      "loss": 1.0357,
      "step": 8308
    },
    {
      "epoch": 0.5989331795574137,
      "grad_norm": 0.11434901314385428,
      "learning_rate": 7.315719323719124e-05,
      "loss": 1.0491,
      "step": 8309
    },
    {
      "epoch": 0.5990052620197506,
      "grad_norm": 0.1726366462562876,
      "learning_rate": 7.31347036645873e-05,
      "loss": 1.0683,
      "step": 8310
    },
    {
      "epoch": 0.5990773444820875,
      "grad_norm": 0.13945116864271226,
      "learning_rate": 7.31122155563824e-05,
      "loss": 1.0831,
      "step": 8311
    },
    {
      "epoch": 0.5991494269444244,
      "grad_norm": 0.12240713830718468,
      "learning_rate": 7.30897289138023e-05,
      "loss": 1.0411,
      "step": 8312
    },
    {
      "epoch": 0.5992215094067613,
      "grad_norm": 0.14049972569671895,
      "learning_rate": 7.30672437380727e-05,
      "loss": 1.0136,
      "step": 8313
    },
    {
      "epoch": 0.5992935918690983,
      "grad_norm": 0.11595902732885346,
      "learning_rate": 7.304476003041928e-05,
      "loss": 1.1129,
      "step": 8314
    },
    {
      "epoch": 0.5993656743314352,
      "grad_norm": 0.1319534493064744,
      "learning_rate": 7.302227779206755e-05,
      "loss": 1.0197,
      "step": 8315
    },
    {
      "epoch": 0.5994377567937721,
      "grad_norm": 0.14799609009367695,
      "learning_rate": 7.299979702424308e-05,
      "loss": 1.0218,
      "step": 8316
    },
    {
      "epoch": 0.599509839256109,
      "grad_norm": 0.432069501823975,
      "learning_rate": 7.29773177281712e-05,
      "loss": 1.1341,
      "step": 8317
    },
    {
      "epoch": 0.5995819217184459,
      "grad_norm": 0.1192354971978527,
      "learning_rate": 7.295483990507727e-05,
      "loss": 1.0179,
      "step": 8318
    },
    {
      "epoch": 0.5996540041807829,
      "grad_norm": 0.25774152726298544,
      "learning_rate": 7.293236355618647e-05,
      "loss": 1.0642,
      "step": 8319
    },
    {
      "epoch": 0.5997260866431198,
      "grad_norm": 0.11472770214644575,
      "learning_rate": 7.290988868272406e-05,
      "loss": 1.0253,
      "step": 8320
    },
    {
      "epoch": 0.5997981691054567,
      "grad_norm": 0.12581641630761,
      "learning_rate": 7.288741528591505e-05,
      "loss": 1.0589,
      "step": 8321
    },
    {
      "epoch": 0.5998702515677936,
      "grad_norm": 0.13991791136179338,
      "learning_rate": 7.286494336698449e-05,
      "loss": 1.0133,
      "step": 8322
    },
    {
      "epoch": 0.5999423340301304,
      "grad_norm": 0.12312475660206247,
      "learning_rate": 7.284247292715721e-05,
      "loss": 1.0123,
      "step": 8323
    },
    {
      "epoch": 0.6000144164924673,
      "grad_norm": 1.5561348560043275,
      "learning_rate": 7.282000396765818e-05,
      "loss": 0.9913,
      "step": 8324
    },
    {
      "epoch": 0.6000864989548043,
      "grad_norm": 0.1400396994545107,
      "learning_rate": 7.279753648971206e-05,
      "loss": 1.0419,
      "step": 8325
    },
    {
      "epoch": 0.6001585814171412,
      "grad_norm": 0.12614510405170445,
      "learning_rate": 7.27750704945436e-05,
      "loss": 1.0808,
      "step": 8326
    },
    {
      "epoch": 0.6002306638794781,
      "grad_norm": 0.16619682128144897,
      "learning_rate": 7.275260598337733e-05,
      "loss": 0.971,
      "step": 8327
    },
    {
      "epoch": 0.600302746341815,
      "grad_norm": 0.16624371140685007,
      "learning_rate": 7.273014295743776e-05,
      "loss": 1.1289,
      "step": 8328
    },
    {
      "epoch": 0.6003748288041519,
      "grad_norm": 1.329930775547965,
      "learning_rate": 7.270768141794942e-05,
      "loss": 1.0553,
      "step": 8329
    },
    {
      "epoch": 0.6004469112664889,
      "grad_norm": 0.2254597773828648,
      "learning_rate": 7.268522136613659e-05,
      "loss": 1.1216,
      "step": 8330
    },
    {
      "epoch": 0.6005189937288258,
      "grad_norm": 0.12020597901189033,
      "learning_rate": 7.266276280322354e-05,
      "loss": 0.948,
      "step": 8331
    },
    {
      "epoch": 0.6005910761911627,
      "grad_norm": 0.1234630503527849,
      "learning_rate": 7.26403057304345e-05,
      "loss": 1.0948,
      "step": 8332
    },
    {
      "epoch": 0.6006631586534996,
      "grad_norm": 0.14201277154257175,
      "learning_rate": 7.261785014899356e-05,
      "loss": 1.0735,
      "step": 8333
    },
    {
      "epoch": 0.6007352411158365,
      "grad_norm": 0.13592851097318773,
      "learning_rate": 7.259539606012478e-05,
      "loss": 1.1216,
      "step": 8334
    },
    {
      "epoch": 0.6008073235781735,
      "grad_norm": 0.12036401229786163,
      "learning_rate": 7.257294346505205e-05,
      "loss": 1.0261,
      "step": 8335
    },
    {
      "epoch": 0.6008794060405104,
      "grad_norm": 0.1419843858617562,
      "learning_rate": 7.255049236499932e-05,
      "loss": 1.0261,
      "step": 8336
    },
    {
      "epoch": 0.6009514885028473,
      "grad_norm": 0.14895402274048994,
      "learning_rate": 7.252804276119023e-05,
      "loss": 1.0517,
      "step": 8337
    },
    {
      "epoch": 0.6010235709651842,
      "grad_norm": 0.12595921546958194,
      "learning_rate": 7.250559465484865e-05,
      "loss": 1.0532,
      "step": 8338
    },
    {
      "epoch": 0.6010956534275211,
      "grad_norm": 0.11921495790805775,
      "learning_rate": 7.248314804719813e-05,
      "loss": 1.0181,
      "step": 8339
    },
    {
      "epoch": 0.601167735889858,
      "grad_norm": 0.10874626624065781,
      "learning_rate": 7.246070293946223e-05,
      "loss": 1.0113,
      "step": 8340
    },
    {
      "epoch": 0.6012398183521949,
      "grad_norm": 0.1478544476437993,
      "learning_rate": 7.243825933286432e-05,
      "loss": 1.0283,
      "step": 8341
    },
    {
      "epoch": 0.6013119008145318,
      "grad_norm": 0.16665335566274359,
      "learning_rate": 7.241581722862791e-05,
      "loss": 1.0381,
      "step": 8342
    },
    {
      "epoch": 0.6013839832768687,
      "grad_norm": 0.11290261963302643,
      "learning_rate": 7.239337662797624e-05,
      "loss": 1.0727,
      "step": 8343
    },
    {
      "epoch": 0.6014560657392056,
      "grad_norm": 0.12822169468939462,
      "learning_rate": 7.237093753213252e-05,
      "loss": 1.0219,
      "step": 8344
    },
    {
      "epoch": 0.6015281482015425,
      "grad_norm": 0.12951951629564132,
      "learning_rate": 7.234849994231982e-05,
      "loss": 0.9662,
      "step": 8345
    },
    {
      "epoch": 0.6016002306638795,
      "grad_norm": 0.14467177492011712,
      "learning_rate": 7.232606385976132e-05,
      "loss": 1.0761,
      "step": 8346
    },
    {
      "epoch": 0.6016723131262164,
      "grad_norm": 0.12008700568727818,
      "learning_rate": 7.230362928567991e-05,
      "loss": 1.0291,
      "step": 8347
    },
    {
      "epoch": 0.6017443955885533,
      "grad_norm": 0.1241272458665965,
      "learning_rate": 7.228119622129848e-05,
      "loss": 1.058,
      "step": 8348
    },
    {
      "epoch": 0.6018164780508902,
      "grad_norm": 0.18158892934613904,
      "learning_rate": 7.225876466783983e-05,
      "loss": 1.0468,
      "step": 8349
    },
    {
      "epoch": 0.6018885605132271,
      "grad_norm": 0.11077595870388729,
      "learning_rate": 7.223633462652665e-05,
      "loss": 1.0423,
      "step": 8350
    },
    {
      "epoch": 0.6019606429755641,
      "grad_norm": 0.10700060307084341,
      "learning_rate": 7.221390609858164e-05,
      "loss": 1.0367,
      "step": 8351
    },
    {
      "epoch": 0.602032725437901,
      "grad_norm": 0.13853306429315157,
      "learning_rate": 7.219147908522735e-05,
      "loss": 1.1075,
      "step": 8352
    },
    {
      "epoch": 0.6021048079002379,
      "grad_norm": 0.12936048168372605,
      "learning_rate": 7.216905358768622e-05,
      "loss": 1.0164,
      "step": 8353
    },
    {
      "epoch": 0.6021768903625748,
      "grad_norm": 0.17369678451582346,
      "learning_rate": 7.214662960718061e-05,
      "loss": 1.0665,
      "step": 8354
    },
    {
      "epoch": 0.6022489728249117,
      "grad_norm": 0.11350972993869501,
      "learning_rate": 7.212420714493291e-05,
      "loss": 1.0664,
      "step": 8355
    },
    {
      "epoch": 0.6023210552872487,
      "grad_norm": 0.13022170696305937,
      "learning_rate": 7.210178620216532e-05,
      "loss": 1.0165,
      "step": 8356
    },
    {
      "epoch": 0.6023931377495855,
      "grad_norm": 0.136100680556367,
      "learning_rate": 7.207936678009996e-05,
      "loss": 1.0595,
      "step": 8357
    },
    {
      "epoch": 0.6024652202119224,
      "grad_norm": 0.12582490616776532,
      "learning_rate": 7.205694887995888e-05,
      "loss": 1.0852,
      "step": 8358
    },
    {
      "epoch": 0.6025373026742593,
      "grad_norm": 0.11882856192325268,
      "learning_rate": 7.203453250296404e-05,
      "loss": 1.1612,
      "step": 8359
    },
    {
      "epoch": 0.6026093851365962,
      "grad_norm": 0.10688892937423629,
      "learning_rate": 7.201211765033743e-05,
      "loss": 0.9791,
      "step": 8360
    },
    {
      "epoch": 0.6026814675989332,
      "grad_norm": 0.10864851273744389,
      "learning_rate": 7.198970432330076e-05,
      "loss": 1.0792,
      "step": 8361
    },
    {
      "epoch": 0.6027535500612701,
      "grad_norm": 0.14199620625201365,
      "learning_rate": 7.196729252307579e-05,
      "loss": 1.0565,
      "step": 8362
    },
    {
      "epoch": 0.602825632523607,
      "grad_norm": 0.11555634283822254,
      "learning_rate": 7.194488225088417e-05,
      "loss": 1.0817,
      "step": 8363
    },
    {
      "epoch": 0.6028977149859439,
      "grad_norm": 0.13413654124549423,
      "learning_rate": 7.192247350794746e-05,
      "loss": 1.0264,
      "step": 8364
    },
    {
      "epoch": 0.6029697974482808,
      "grad_norm": 0.12032643637020954,
      "learning_rate": 7.190006629548714e-05,
      "loss": 0.98,
      "step": 8365
    },
    {
      "epoch": 0.6030418799106178,
      "grad_norm": 0.15481750013107054,
      "learning_rate": 7.187766061472459e-05,
      "loss": 1.1117,
      "step": 8366
    },
    {
      "epoch": 0.6031139623729547,
      "grad_norm": 0.21663765184621933,
      "learning_rate": 7.18552564668811e-05,
      "loss": 0.9322,
      "step": 8367
    },
    {
      "epoch": 0.6031860448352916,
      "grad_norm": 0.14662354452320947,
      "learning_rate": 7.183285385317795e-05,
      "loss": 1.0984,
      "step": 8368
    },
    {
      "epoch": 0.6032581272976285,
      "grad_norm": 0.11910951790754501,
      "learning_rate": 7.181045277483625e-05,
      "loss": 1.0033,
      "step": 8369
    },
    {
      "epoch": 0.6033302097599654,
      "grad_norm": 0.10908357888182556,
      "learning_rate": 7.178805323307707e-05,
      "loss": 1.0564,
      "step": 8370
    },
    {
      "epoch": 0.6034022922223023,
      "grad_norm": 0.11316603254207754,
      "learning_rate": 7.176565522912138e-05,
      "loss": 0.9965,
      "step": 8371
    },
    {
      "epoch": 0.6034743746846393,
      "grad_norm": 0.14017428895016681,
      "learning_rate": 7.174325876419006e-05,
      "loss": 0.9589,
      "step": 8372
    },
    {
      "epoch": 0.6035464571469762,
      "grad_norm": 0.11494080160758592,
      "learning_rate": 7.172086383950393e-05,
      "loss": 1.007,
      "step": 8373
    },
    {
      "epoch": 0.6036185396093131,
      "grad_norm": 0.10585456474997687,
      "learning_rate": 7.16984704562837e-05,
      "loss": 0.9935,
      "step": 8374
    },
    {
      "epoch": 0.6036906220716499,
      "grad_norm": 0.21737599717624329,
      "learning_rate": 7.167607861575007e-05,
      "loss": 1.0449,
      "step": 8375
    },
    {
      "epoch": 0.6037627045339868,
      "grad_norm": 0.1240334699061205,
      "learning_rate": 7.16536883191235e-05,
      "loss": 1.0479,
      "step": 8376
    },
    {
      "epoch": 0.6038347869963238,
      "grad_norm": 0.11270620985905247,
      "learning_rate": 7.163129956762454e-05,
      "loss": 1.0203,
      "step": 8377
    },
    {
      "epoch": 0.6039068694586607,
      "grad_norm": 0.13798293767261735,
      "learning_rate": 7.160891236247353e-05,
      "loss": 0.9992,
      "step": 8378
    },
    {
      "epoch": 0.6039789519209976,
      "grad_norm": 0.14721210820935232,
      "learning_rate": 7.15865267048908e-05,
      "loss": 0.9684,
      "step": 8379
    },
    {
      "epoch": 0.6040510343833345,
      "grad_norm": 0.18107991637026508,
      "learning_rate": 7.156414259609657e-05,
      "loss": 1.0582,
      "step": 8380
    },
    {
      "epoch": 0.6041231168456714,
      "grad_norm": 0.1900670938975177,
      "learning_rate": 7.154176003731092e-05,
      "loss": 1.0492,
      "step": 8381
    },
    {
      "epoch": 0.6041951993080084,
      "grad_norm": 0.11831924563857568,
      "learning_rate": 7.151937902975398e-05,
      "loss": 1.0191,
      "step": 8382
    },
    {
      "epoch": 0.6042672817703453,
      "grad_norm": 0.13279726361028862,
      "learning_rate": 7.149699957464568e-05,
      "loss": 1.0458,
      "step": 8383
    },
    {
      "epoch": 0.6043393642326822,
      "grad_norm": 0.17278459085899114,
      "learning_rate": 7.147462167320589e-05,
      "loss": 1.0869,
      "step": 8384
    },
    {
      "epoch": 0.6044114466950191,
      "grad_norm": 0.12003662007272352,
      "learning_rate": 7.14522453266544e-05,
      "loss": 1.0665,
      "step": 8385
    },
    {
      "epoch": 0.604483529157356,
      "grad_norm": 0.13019920370509866,
      "learning_rate": 7.142987053621094e-05,
      "loss": 1.1255,
      "step": 8386
    },
    {
      "epoch": 0.604555611619693,
      "grad_norm": 0.12524909832652156,
      "learning_rate": 7.140749730309516e-05,
      "loss": 1.0918,
      "step": 8387
    },
    {
      "epoch": 0.6046276940820299,
      "grad_norm": 0.1572623642887794,
      "learning_rate": 7.138512562852655e-05,
      "loss": 1.1005,
      "step": 8388
    },
    {
      "epoch": 0.6046997765443668,
      "grad_norm": 0.12548780146375763,
      "learning_rate": 7.136275551372458e-05,
      "loss": 0.9929,
      "step": 8389
    },
    {
      "epoch": 0.6047718590067037,
      "grad_norm": 0.14996221330524562,
      "learning_rate": 7.134038695990863e-05,
      "loss": 1.052,
      "step": 8390
    },
    {
      "epoch": 0.6048439414690406,
      "grad_norm": 0.1652825161212299,
      "learning_rate": 7.131801996829802e-05,
      "loss": 0.9919,
      "step": 8391
    },
    {
      "epoch": 0.6049160239313774,
      "grad_norm": 0.33785724619435453,
      "learning_rate": 7.129565454011189e-05,
      "loss": 1.0537,
      "step": 8392
    },
    {
      "epoch": 0.6049881063937144,
      "grad_norm": 0.1296744803151949,
      "learning_rate": 7.127329067656941e-05,
      "loss": 1.0646,
      "step": 8393
    },
    {
      "epoch": 0.6050601888560513,
      "grad_norm": 0.13605001988298351,
      "learning_rate": 7.125092837888952e-05,
      "loss": 1.0661,
      "step": 8394
    },
    {
      "epoch": 0.6051322713183882,
      "grad_norm": 0.1282508153420755,
      "learning_rate": 7.12285676482913e-05,
      "loss": 1.0608,
      "step": 8395
    },
    {
      "epoch": 0.6052043537807251,
      "grad_norm": 0.15192140619267086,
      "learning_rate": 7.12062084859935e-05,
      "loss": 1.0636,
      "step": 8396
    },
    {
      "epoch": 0.605276436243062,
      "grad_norm": 0.13529199084137591,
      "learning_rate": 7.118385089321498e-05,
      "loss": 1.0247,
      "step": 8397
    },
    {
      "epoch": 0.605348518705399,
      "grad_norm": 0.11558863000120691,
      "learning_rate": 7.116149487117432e-05,
      "loss": 1.0955,
      "step": 8398
    },
    {
      "epoch": 0.6054206011677359,
      "grad_norm": 0.15261338595995053,
      "learning_rate": 7.113914042109023e-05,
      "loss": 1.0147,
      "step": 8399
    },
    {
      "epoch": 0.6054926836300728,
      "grad_norm": 0.11565280960351405,
      "learning_rate": 7.111678754418117e-05,
      "loss": 1.0464,
      "step": 8400
    },
    {
      "epoch": 0.6055647660924097,
      "grad_norm": 0.12792768255521572,
      "learning_rate": 7.10944362416656e-05,
      "loss": 1.008,
      "step": 8401
    },
    {
      "epoch": 0.6056368485547466,
      "grad_norm": 0.10256267867584326,
      "learning_rate": 7.107208651476185e-05,
      "loss": 1.0267,
      "step": 8402
    },
    {
      "epoch": 0.6057089310170836,
      "grad_norm": 0.13348954610398528,
      "learning_rate": 7.104973836468815e-05,
      "loss": 1.062,
      "step": 8403
    },
    {
      "epoch": 0.6057810134794205,
      "grad_norm": 0.12894210980936477,
      "learning_rate": 7.102739179266271e-05,
      "loss": 1.0197,
      "step": 8404
    },
    {
      "epoch": 0.6058530959417574,
      "grad_norm": 0.1251277811655742,
      "learning_rate": 7.100504679990365e-05,
      "loss": 1.1069,
      "step": 8405
    },
    {
      "epoch": 0.6059251784040943,
      "grad_norm": 0.11389845345181357,
      "learning_rate": 7.098270338762891e-05,
      "loss": 1.0157,
      "step": 8406
    },
    {
      "epoch": 0.6059972608664312,
      "grad_norm": 0.11894892651132767,
      "learning_rate": 7.096036155705643e-05,
      "loss": 1.0364,
      "step": 8407
    },
    {
      "epoch": 0.6060693433287682,
      "grad_norm": 0.17509129173035648,
      "learning_rate": 7.093802130940403e-05,
      "loss": 1.0434,
      "step": 8408
    },
    {
      "epoch": 0.606141425791105,
      "grad_norm": 0.11539585912053306,
      "learning_rate": 7.091568264588952e-05,
      "loss": 1.0426,
      "step": 8409
    },
    {
      "epoch": 0.6062135082534419,
      "grad_norm": 0.1337776144602306,
      "learning_rate": 7.089334556773046e-05,
      "loss": 1.0985,
      "step": 8410
    },
    {
      "epoch": 0.6062855907157788,
      "grad_norm": 0.11080610660559884,
      "learning_rate": 7.087101007614443e-05,
      "loss": 1.1,
      "step": 8411
    },
    {
      "epoch": 0.6063576731781157,
      "grad_norm": 0.13322621356637288,
      "learning_rate": 7.084867617234898e-05,
      "loss": 1.049,
      "step": 8412
    },
    {
      "epoch": 0.6064297556404527,
      "grad_norm": 0.20619769555664352,
      "learning_rate": 7.08263438575615e-05,
      "loss": 1.0208,
      "step": 8413
    },
    {
      "epoch": 0.6065018381027896,
      "grad_norm": 0.17583133599960352,
      "learning_rate": 7.080401313299925e-05,
      "loss": 1.011,
      "step": 8414
    },
    {
      "epoch": 0.6065739205651265,
      "grad_norm": 0.11450918422495927,
      "learning_rate": 7.07816839998795e-05,
      "loss": 1.0451,
      "step": 8415
    },
    {
      "epoch": 0.6066460030274634,
      "grad_norm": 0.12436855649780246,
      "learning_rate": 7.075935645941932e-05,
      "loss": 1.0104,
      "step": 8416
    },
    {
      "epoch": 0.6067180854898003,
      "grad_norm": 0.16498309904962546,
      "learning_rate": 7.073703051283584e-05,
      "loss": 1.0934,
      "step": 8417
    },
    {
      "epoch": 0.6067901679521372,
      "grad_norm": 0.12032498917372039,
      "learning_rate": 7.071470616134599e-05,
      "loss": 1.0948,
      "step": 8418
    },
    {
      "epoch": 0.6068622504144742,
      "grad_norm": 0.13200964203309795,
      "learning_rate": 7.069238340616666e-05,
      "loss": 1.0575,
      "step": 8419
    },
    {
      "epoch": 0.6069343328768111,
      "grad_norm": 0.11691447693453223,
      "learning_rate": 7.067006224851456e-05,
      "loss": 1.0058,
      "step": 8420
    },
    {
      "epoch": 0.607006415339148,
      "grad_norm": 0.17624950044025892,
      "learning_rate": 7.064774268960653e-05,
      "loss": 1.0295,
      "step": 8421
    },
    {
      "epoch": 0.6070784978014849,
      "grad_norm": 0.11363927782935236,
      "learning_rate": 7.06254247306591e-05,
      "loss": 1.0746,
      "step": 8422
    },
    {
      "epoch": 0.6071505802638218,
      "grad_norm": 0.14325833932490337,
      "learning_rate": 7.060310837288884e-05,
      "loss": 1.0756,
      "step": 8423
    },
    {
      "epoch": 0.6072226627261588,
      "grad_norm": 0.1367436992385731,
      "learning_rate": 7.058079361751213e-05,
      "loss": 1.1353,
      "step": 8424
    },
    {
      "epoch": 0.6072947451884957,
      "grad_norm": 0.18469230698300776,
      "learning_rate": 7.055848046574534e-05,
      "loss": 1.1171,
      "step": 8425
    },
    {
      "epoch": 0.6073668276508325,
      "grad_norm": 0.1390017756800717,
      "learning_rate": 7.05361689188048e-05,
      "loss": 1.0499,
      "step": 8426
    },
    {
      "epoch": 0.6074389101131694,
      "grad_norm": 0.11950033110578857,
      "learning_rate": 7.051385897790663e-05,
      "loss": 1.0258,
      "step": 8427
    },
    {
      "epoch": 0.6075109925755063,
      "grad_norm": 0.14379174491999394,
      "learning_rate": 7.049155064426693e-05,
      "loss": 0.9626,
      "step": 8428
    },
    {
      "epoch": 0.6075830750378433,
      "grad_norm": 0.11264392286032589,
      "learning_rate": 7.04692439191017e-05,
      "loss": 1.0366,
      "step": 8429
    },
    {
      "epoch": 0.6076551575001802,
      "grad_norm": 0.13201106181848854,
      "learning_rate": 7.044693880362687e-05,
      "loss": 1.0689,
      "step": 8430
    },
    {
      "epoch": 0.6077272399625171,
      "grad_norm": 0.13236851747449413,
      "learning_rate": 7.042463529905829e-05,
      "loss": 1.0084,
      "step": 8431
    },
    {
      "epoch": 0.607799322424854,
      "grad_norm": 0.1409315784971688,
      "learning_rate": 7.040233340661167e-05,
      "loss": 1.069,
      "step": 8432
    },
    {
      "epoch": 0.6078714048871909,
      "grad_norm": 0.1351919090173791,
      "learning_rate": 7.038003312750263e-05,
      "loss": 1.0128,
      "step": 8433
    },
    {
      "epoch": 0.6079434873495279,
      "grad_norm": 0.13469499657875464,
      "learning_rate": 7.03577344629468e-05,
      "loss": 1.0365,
      "step": 8434
    },
    {
      "epoch": 0.6080155698118648,
      "grad_norm": 0.11668354905526047,
      "learning_rate": 7.033543741415964e-05,
      "loss": 0.9986,
      "step": 8435
    },
    {
      "epoch": 0.6080876522742017,
      "grad_norm": 0.11031615397529848,
      "learning_rate": 7.031314198235652e-05,
      "loss": 1.0748,
      "step": 8436
    },
    {
      "epoch": 0.6081597347365386,
      "grad_norm": 0.20177187286692738,
      "learning_rate": 7.029084816875275e-05,
      "loss": 1.0227,
      "step": 8437
    },
    {
      "epoch": 0.6082318171988755,
      "grad_norm": 0.1250031161008353,
      "learning_rate": 7.02685559745635e-05,
      "loss": 0.981,
      "step": 8438
    },
    {
      "epoch": 0.6083038996612125,
      "grad_norm": 0.11986708687167531,
      "learning_rate": 7.024626540100401e-05,
      "loss": 1.0942,
      "step": 8439
    },
    {
      "epoch": 0.6083759821235494,
      "grad_norm": 0.16503487324338292,
      "learning_rate": 7.022397644928919e-05,
      "loss": 1.0168,
      "step": 8440
    },
    {
      "epoch": 0.6084480645858863,
      "grad_norm": 0.13272256060082085,
      "learning_rate": 7.020168912063405e-05,
      "loss": 1.0616,
      "step": 8441
    },
    {
      "epoch": 0.6085201470482232,
      "grad_norm": 0.12339987082045074,
      "learning_rate": 7.01794034162534e-05,
      "loss": 1.0474,
      "step": 8442
    },
    {
      "epoch": 0.6085922295105601,
      "grad_norm": 0.15965620339935593,
      "learning_rate": 7.015711933736211e-05,
      "loss": 1.0615,
      "step": 8443
    },
    {
      "epoch": 0.6086643119728969,
      "grad_norm": 0.1516296757254509,
      "learning_rate": 7.013483688517476e-05,
      "loss": 1.0533,
      "step": 8444
    },
    {
      "epoch": 0.6087363944352339,
      "grad_norm": 0.11334856813656434,
      "learning_rate": 7.011255606090599e-05,
      "loss": 1.0638,
      "step": 8445
    },
    {
      "epoch": 0.6088084768975708,
      "grad_norm": 0.12065474435708838,
      "learning_rate": 7.009027686577032e-05,
      "loss": 0.9323,
      "step": 8446
    },
    {
      "epoch": 0.6088805593599077,
      "grad_norm": 0.17319880112185723,
      "learning_rate": 7.00679993009821e-05,
      "loss": 1.0166,
      "step": 8447
    },
    {
      "epoch": 0.6089526418222446,
      "grad_norm": 0.39708059438028015,
      "learning_rate": 7.00457233677557e-05,
      "loss": 1.1219,
      "step": 8448
    },
    {
      "epoch": 0.6090247242845815,
      "grad_norm": 0.10406902052304981,
      "learning_rate": 7.002344906730536e-05,
      "loss": 0.9231,
      "step": 8449
    },
    {
      "epoch": 0.6090968067469185,
      "grad_norm": 0.18124728460189798,
      "learning_rate": 7.000117640084526e-05,
      "loss": 0.9993,
      "step": 8450
    },
    {
      "epoch": 0.6091688892092554,
      "grad_norm": 0.208702693727799,
      "learning_rate": 6.997890536958937e-05,
      "loss": 1.0656,
      "step": 8451
    },
    {
      "epoch": 0.6092409716715923,
      "grad_norm": 0.11064955376149993,
      "learning_rate": 6.995663597475174e-05,
      "loss": 1.0121,
      "step": 8452
    },
    {
      "epoch": 0.6093130541339292,
      "grad_norm": 0.12328832206500949,
      "learning_rate": 6.993436821754621e-05,
      "loss": 1.0464,
      "step": 8453
    },
    {
      "epoch": 0.6093851365962661,
      "grad_norm": 0.11669498176284765,
      "learning_rate": 6.991210209918663e-05,
      "loss": 0.9742,
      "step": 8454
    },
    {
      "epoch": 0.6094572190586031,
      "grad_norm": 0.12248121728803071,
      "learning_rate": 6.988983762088661e-05,
      "loss": 1.0111,
      "step": 8455
    },
    {
      "epoch": 0.60952930152094,
      "grad_norm": 0.12047370963874905,
      "learning_rate": 6.986757478385983e-05,
      "loss": 1.0835,
      "step": 8456
    },
    {
      "epoch": 0.6096013839832769,
      "grad_norm": 0.16644494417918684,
      "learning_rate": 6.98453135893198e-05,
      "loss": 1.1045,
      "step": 8457
    },
    {
      "epoch": 0.6096734664456138,
      "grad_norm": 0.12003830142473694,
      "learning_rate": 6.982305403847997e-05,
      "loss": 1.0534,
      "step": 8458
    },
    {
      "epoch": 0.6097455489079507,
      "grad_norm": 0.21042342059517752,
      "learning_rate": 6.980079613255364e-05,
      "loss": 1.11,
      "step": 8459
    },
    {
      "epoch": 0.6098176313702877,
      "grad_norm": 0.1229009724680351,
      "learning_rate": 6.977853987275408e-05,
      "loss": 1.0456,
      "step": 8460
    },
    {
      "epoch": 0.6098897138326245,
      "grad_norm": 0.11021934498212081,
      "learning_rate": 6.975628526029447e-05,
      "loss": 0.9902,
      "step": 8461
    },
    {
      "epoch": 0.6099617962949614,
      "grad_norm": 0.11504482141359361,
      "learning_rate": 6.973403229638793e-05,
      "loss": 0.992,
      "step": 8462
    },
    {
      "epoch": 0.6100338787572983,
      "grad_norm": 0.13111898424557045,
      "learning_rate": 6.971178098224735e-05,
      "loss": 1.0597,
      "step": 8463
    },
    {
      "epoch": 0.6101059612196352,
      "grad_norm": 0.13189778716945427,
      "learning_rate": 6.968953131908567e-05,
      "loss": 1.104,
      "step": 8464
    },
    {
      "epoch": 0.6101780436819721,
      "grad_norm": 0.12078394814992982,
      "learning_rate": 6.96672833081157e-05,
      "loss": 0.9953,
      "step": 8465
    },
    {
      "epoch": 0.6102501261443091,
      "grad_norm": 0.1273258549769993,
      "learning_rate": 6.964503695055019e-05,
      "loss": 0.983,
      "step": 8466
    },
    {
      "epoch": 0.610322208606646,
      "grad_norm": 0.12584808322588792,
      "learning_rate": 6.962279224760172e-05,
      "loss": 1.0732,
      "step": 8467
    },
    {
      "epoch": 0.6103942910689829,
      "grad_norm": 0.1173122826872515,
      "learning_rate": 6.960054920048282e-05,
      "loss": 1.0266,
      "step": 8468
    },
    {
      "epoch": 0.6104663735313198,
      "grad_norm": 0.11100002478192059,
      "learning_rate": 6.957830781040591e-05,
      "loss": 1.0614,
      "step": 8469
    },
    {
      "epoch": 0.6105384559936567,
      "grad_norm": 0.11970165280359431,
      "learning_rate": 6.955606807858343e-05,
      "loss": 1.048,
      "step": 8470
    },
    {
      "epoch": 0.6106105384559937,
      "grad_norm": 0.13916958609701033,
      "learning_rate": 6.953383000622761e-05,
      "loss": 1.0116,
      "step": 8471
    },
    {
      "epoch": 0.6106826209183306,
      "grad_norm": 0.11566102522776008,
      "learning_rate": 6.951159359455062e-05,
      "loss": 1.0944,
      "step": 8472
    },
    {
      "epoch": 0.6107547033806675,
      "grad_norm": 0.14304203061255497,
      "learning_rate": 6.948935884476446e-05,
      "loss": 1.0856,
      "step": 8473
    },
    {
      "epoch": 0.6108267858430044,
      "grad_norm": 0.21936139449976655,
      "learning_rate": 6.946712575808128e-05,
      "loss": 1.0664,
      "step": 8474
    },
    {
      "epoch": 0.6108988683053413,
      "grad_norm": 0.13033206406018305,
      "learning_rate": 6.944489433571286e-05,
      "loss": 1.0756,
      "step": 8475
    },
    {
      "epoch": 0.6109709507676783,
      "grad_norm": 0.11259208692086131,
      "learning_rate": 6.942266457887109e-05,
      "loss": 1.0396,
      "step": 8476
    },
    {
      "epoch": 0.6110430332300152,
      "grad_norm": 0.18117176629237092,
      "learning_rate": 6.940043648876758e-05,
      "loss": 1.0308,
      "step": 8477
    },
    {
      "epoch": 0.611115115692352,
      "grad_norm": 0.13314528028492492,
      "learning_rate": 6.93782100666141e-05,
      "loss": 1.0494,
      "step": 8478
    },
    {
      "epoch": 0.6111871981546889,
      "grad_norm": 0.13169856585881287,
      "learning_rate": 6.93559853136221e-05,
      "loss": 1.065,
      "step": 8479
    },
    {
      "epoch": 0.6112592806170258,
      "grad_norm": 0.12130199384852204,
      "learning_rate": 6.933376223100307e-05,
      "loss": 1.1161,
      "step": 8480
    },
    {
      "epoch": 0.6113313630793628,
      "grad_norm": 0.12158674223202734,
      "learning_rate": 6.931154081996833e-05,
      "loss": 1.0466,
      "step": 8481
    },
    {
      "epoch": 0.6114034455416997,
      "grad_norm": 0.14299001709139247,
      "learning_rate": 6.928932108172914e-05,
      "loss": 1.0337,
      "step": 8482
    },
    {
      "epoch": 0.6114755280040366,
      "grad_norm": 0.12904379474992284,
      "learning_rate": 6.926710301749673e-05,
      "loss": 1.0976,
      "step": 8483
    },
    {
      "epoch": 0.6115476104663735,
      "grad_norm": 0.12687123472821588,
      "learning_rate": 6.924488662848214e-05,
      "loss": 1.0246,
      "step": 8484
    },
    {
      "epoch": 0.6116196929287104,
      "grad_norm": 0.25436796336970274,
      "learning_rate": 6.922267191589637e-05,
      "loss": 0.9943,
      "step": 8485
    },
    {
      "epoch": 0.6116917753910474,
      "grad_norm": 0.1558787464171885,
      "learning_rate": 6.92004588809503e-05,
      "loss": 1.0001,
      "step": 8486
    },
    {
      "epoch": 0.6117638578533843,
      "grad_norm": 0.13703039502427622,
      "learning_rate": 6.917824752485477e-05,
      "loss": 1.0727,
      "step": 8487
    },
    {
      "epoch": 0.6118359403157212,
      "grad_norm": 0.11593117880251269,
      "learning_rate": 6.915603784882054e-05,
      "loss": 1.0826,
      "step": 8488
    },
    {
      "epoch": 0.6119080227780581,
      "grad_norm": 0.2307200748105448,
      "learning_rate": 6.913382985405814e-05,
      "loss": 1.0765,
      "step": 8489
    },
    {
      "epoch": 0.611980105240395,
      "grad_norm": 0.19533432488059857,
      "learning_rate": 6.911162354177818e-05,
      "loss": 1.0338,
      "step": 8490
    },
    {
      "epoch": 0.612052187702732,
      "grad_norm": 0.12470104375732265,
      "learning_rate": 6.9089418913191e-05,
      "loss": 1.0171,
      "step": 8491
    },
    {
      "epoch": 0.6121242701650689,
      "grad_norm": 0.12633134250806985,
      "learning_rate": 6.90672159695071e-05,
      "loss": 1.1248,
      "step": 8492
    },
    {
      "epoch": 0.6121963526274058,
      "grad_norm": 0.12409561643037477,
      "learning_rate": 6.904501471193663e-05,
      "loss": 1.085,
      "step": 8493
    },
    {
      "epoch": 0.6122684350897427,
      "grad_norm": 0.13262598525308064,
      "learning_rate": 6.902281514168983e-05,
      "loss": 1.0486,
      "step": 8494
    },
    {
      "epoch": 0.6123405175520796,
      "grad_norm": 0.2542367583376633,
      "learning_rate": 6.900061725997664e-05,
      "loss": 0.9893,
      "step": 8495
    },
    {
      "epoch": 0.6124126000144164,
      "grad_norm": 0.2166695869299247,
      "learning_rate": 6.897842106800722e-05,
      "loss": 1.0349,
      "step": 8496
    },
    {
      "epoch": 0.6124846824767534,
      "grad_norm": 0.11055727428471315,
      "learning_rate": 6.895622656699135e-05,
      "loss": 1.0016,
      "step": 8497
    },
    {
      "epoch": 0.6125567649390903,
      "grad_norm": 0.165768965759127,
      "learning_rate": 6.893403375813888e-05,
      "loss": 1.0746,
      "step": 8498
    },
    {
      "epoch": 0.6126288474014272,
      "grad_norm": 0.11315806811359701,
      "learning_rate": 6.891184264265944e-05,
      "loss": 0.9562,
      "step": 8499
    },
    {
      "epoch": 0.6127009298637641,
      "grad_norm": 0.11541161487119476,
      "learning_rate": 6.888965322176276e-05,
      "loss": 1.1001,
      "step": 8500
    },
    {
      "epoch": 0.612773012326101,
      "grad_norm": 0.1946073429629242,
      "learning_rate": 6.886746549665827e-05,
      "loss": 1.0239,
      "step": 8501
    },
    {
      "epoch": 0.612845094788438,
      "grad_norm": 0.12082459921758451,
      "learning_rate": 6.884527946855546e-05,
      "loss": 1.0036,
      "step": 8502
    },
    {
      "epoch": 0.6129171772507749,
      "grad_norm": 0.14722847549943666,
      "learning_rate": 6.88230951386636e-05,
      "loss": 1.0307,
      "step": 8503
    },
    {
      "epoch": 0.6129892597131118,
      "grad_norm": 0.17396047979307847,
      "learning_rate": 6.880091250819197e-05,
      "loss": 0.9749,
      "step": 8504
    },
    {
      "epoch": 0.6130613421754487,
      "grad_norm": 0.11774931025286102,
      "learning_rate": 6.877873157834974e-05,
      "loss": 1.0408,
      "step": 8505
    },
    {
      "epoch": 0.6131334246377856,
      "grad_norm": 0.11910934581061734,
      "learning_rate": 6.875655235034597e-05,
      "loss": 1.0335,
      "step": 8506
    },
    {
      "epoch": 0.6132055071001226,
      "grad_norm": 0.2011122985051062,
      "learning_rate": 6.873437482538958e-05,
      "loss": 1.0211,
      "step": 8507
    },
    {
      "epoch": 0.6132775895624595,
      "grad_norm": 0.1160066885368462,
      "learning_rate": 6.871219900468947e-05,
      "loss": 1.0679,
      "step": 8508
    },
    {
      "epoch": 0.6133496720247964,
      "grad_norm": 0.15421200350848385,
      "learning_rate": 6.869002488945442e-05,
      "loss": 0.9874,
      "step": 8509
    },
    {
      "epoch": 0.6134217544871333,
      "grad_norm": 0.11952333873857249,
      "learning_rate": 6.866785248089315e-05,
      "loss": 1.1077,
      "step": 8510
    },
    {
      "epoch": 0.6134938369494702,
      "grad_norm": 0.14144496561973285,
      "learning_rate": 6.864568178021423e-05,
      "loss": 1.0506,
      "step": 8511
    },
    {
      "epoch": 0.6135659194118072,
      "grad_norm": 0.1220724283844445,
      "learning_rate": 6.862351278862613e-05,
      "loss": 1.0412,
      "step": 8512
    },
    {
      "epoch": 0.613638001874144,
      "grad_norm": 0.1221560759845794,
      "learning_rate": 6.860134550733726e-05,
      "loss": 1.049,
      "step": 8513
    },
    {
      "epoch": 0.6137100843364809,
      "grad_norm": 0.18523014480022412,
      "learning_rate": 6.857917993755602e-05,
      "loss": 1.0841,
      "step": 8514
    },
    {
      "epoch": 0.6137821667988178,
      "grad_norm": 0.11832703385031886,
      "learning_rate": 6.855701608049054e-05,
      "loss": 1.0454,
      "step": 8515
    },
    {
      "epoch": 0.6138542492611547,
      "grad_norm": 0.12391677861789503,
      "learning_rate": 6.853485393734899e-05,
      "loss": 1.0719,
      "step": 8516
    },
    {
      "epoch": 0.6139263317234916,
      "grad_norm": 0.11942384953883577,
      "learning_rate": 6.851269350933936e-05,
      "loss": 1.0732,
      "step": 8517
    },
    {
      "epoch": 0.6139984141858286,
      "grad_norm": 0.15791158267229208,
      "learning_rate": 6.849053479766968e-05,
      "loss": 1.1457,
      "step": 8518
    },
    {
      "epoch": 0.6140704966481655,
      "grad_norm": 0.15765465023512784,
      "learning_rate": 6.846837780354772e-05,
      "loss": 1.0421,
      "step": 8519
    },
    {
      "epoch": 0.6141425791105024,
      "grad_norm": 0.11884228719748244,
      "learning_rate": 6.844622252818127e-05,
      "loss": 1.0408,
      "step": 8520
    },
    {
      "epoch": 0.6142146615728393,
      "grad_norm": 0.15106722982170612,
      "learning_rate": 6.842406897277795e-05,
      "loss": 1.0597,
      "step": 8521
    },
    {
      "epoch": 0.6142867440351762,
      "grad_norm": 0.13014738939359471,
      "learning_rate": 6.84019171385454e-05,
      "loss": 1.053,
      "step": 8522
    },
    {
      "epoch": 0.6143588264975132,
      "grad_norm": 0.1357973203604376,
      "learning_rate": 6.837976702669103e-05,
      "loss": 1.0865,
      "step": 8523
    },
    {
      "epoch": 0.6144309089598501,
      "grad_norm": 0.12411663315238824,
      "learning_rate": 6.835761863842224e-05,
      "loss": 1.0907,
      "step": 8524
    },
    {
      "epoch": 0.614502991422187,
      "grad_norm": 0.15525857989875175,
      "learning_rate": 6.833547197494632e-05,
      "loss": 1.1128,
      "step": 8525
    },
    {
      "epoch": 0.6145750738845239,
      "grad_norm": 0.1091877346102747,
      "learning_rate": 6.831332703747044e-05,
      "loss": 1.0252,
      "step": 8526
    },
    {
      "epoch": 0.6146471563468608,
      "grad_norm": 0.1153882987610181,
      "learning_rate": 6.829118382720173e-05,
      "loss": 0.9571,
      "step": 8527
    },
    {
      "epoch": 0.6147192388091978,
      "grad_norm": 0.12366063561458734,
      "learning_rate": 6.826904234534717e-05,
      "loss": 0.9969,
      "step": 8528
    },
    {
      "epoch": 0.6147913212715347,
      "grad_norm": 0.1126840782837023,
      "learning_rate": 6.824690259311369e-05,
      "loss": 1.0748,
      "step": 8529
    },
    {
      "epoch": 0.6148634037338715,
      "grad_norm": 0.15889321582927185,
      "learning_rate": 6.822476457170806e-05,
      "loss": 1.0417,
      "step": 8530
    },
    {
      "epoch": 0.6149354861962084,
      "grad_norm": 0.13236159232668926,
      "learning_rate": 6.820262828233704e-05,
      "loss": 0.9982,
      "step": 8531
    },
    {
      "epoch": 0.6150075686585453,
      "grad_norm": 0.13914083692218115,
      "learning_rate": 6.818049372620726e-05,
      "loss": 1.0459,
      "step": 8532
    },
    {
      "epoch": 0.6150796511208823,
      "grad_norm": 0.33841490533775037,
      "learning_rate": 6.815836090452525e-05,
      "loss": 1.0377,
      "step": 8533
    },
    {
      "epoch": 0.6151517335832192,
      "grad_norm": 0.14042818683997668,
      "learning_rate": 6.813622981849742e-05,
      "loss": 1.1127,
      "step": 8534
    },
    {
      "epoch": 0.6152238160455561,
      "grad_norm": 0.12985387727734737,
      "learning_rate": 6.811410046933009e-05,
      "loss": 1.0875,
      "step": 8535
    },
    {
      "epoch": 0.615295898507893,
      "grad_norm": 0.1487118207668958,
      "learning_rate": 6.809197285822959e-05,
      "loss": 1.0166,
      "step": 8536
    },
    {
      "epoch": 0.6153679809702299,
      "grad_norm": 0.11966938769112956,
      "learning_rate": 6.806984698640202e-05,
      "loss": 1.073,
      "step": 8537
    },
    {
      "epoch": 0.6154400634325669,
      "grad_norm": 0.1250032154785379,
      "learning_rate": 6.804772285505344e-05,
      "loss": 1.0212,
      "step": 8538
    },
    {
      "epoch": 0.6155121458949038,
      "grad_norm": 0.145781678439563,
      "learning_rate": 6.80256004653898e-05,
      "loss": 1.0563,
      "step": 8539
    },
    {
      "epoch": 0.6155842283572407,
      "grad_norm": 0.12411641769946866,
      "learning_rate": 6.800347981861699e-05,
      "loss": 1.0286,
      "step": 8540
    },
    {
      "epoch": 0.6156563108195776,
      "grad_norm": 0.11792478427115041,
      "learning_rate": 6.798136091594081e-05,
      "loss": 1.0161,
      "step": 8541
    },
    {
      "epoch": 0.6157283932819145,
      "grad_norm": 0.11866897545909706,
      "learning_rate": 6.795924375856688e-05,
      "loss": 1.0073,
      "step": 8542
    },
    {
      "epoch": 0.6158004757442515,
      "grad_norm": 0.17506720841339765,
      "learning_rate": 6.793712834770079e-05,
      "loss": 1.0539,
      "step": 8543
    },
    {
      "epoch": 0.6158725582065884,
      "grad_norm": 0.10824987285024004,
      "learning_rate": 6.791501468454808e-05,
      "loss": 1.089,
      "step": 8544
    },
    {
      "epoch": 0.6159446406689253,
      "grad_norm": 0.12172221189933924,
      "learning_rate": 6.789290277031413e-05,
      "loss": 1.0711,
      "step": 8545
    },
    {
      "epoch": 0.6160167231312622,
      "grad_norm": 1.262263951621168,
      "learning_rate": 6.787079260620418e-05,
      "loss": 1.0649,
      "step": 8546
    },
    {
      "epoch": 0.6160888055935991,
      "grad_norm": 0.11971863277776493,
      "learning_rate": 6.784868419342349e-05,
      "loss": 1.041,
      "step": 8547
    },
    {
      "epoch": 0.6161608880559359,
      "grad_norm": 0.14406450816154456,
      "learning_rate": 6.782657753317711e-05,
      "loss": 1.1128,
      "step": 8548
    },
    {
      "epoch": 0.6162329705182729,
      "grad_norm": 0.14154193070534457,
      "learning_rate": 6.780447262667012e-05,
      "loss": 0.9853,
      "step": 8549
    },
    {
      "epoch": 0.6163050529806098,
      "grad_norm": 0.2372471410338803,
      "learning_rate": 6.778236947510739e-05,
      "loss": 1.1063,
      "step": 8550
    },
    {
      "epoch": 0.6163771354429467,
      "grad_norm": 0.13665944694770768,
      "learning_rate": 6.776026807969378e-05,
      "loss": 1.0774,
      "step": 8551
    },
    {
      "epoch": 0.6164492179052836,
      "grad_norm": 0.11674893138829913,
      "learning_rate": 6.773816844163393e-05,
      "loss": 1.0908,
      "step": 8552
    },
    {
      "epoch": 0.6165213003676205,
      "grad_norm": 0.17311453783345138,
      "learning_rate": 6.771607056213257e-05,
      "loss": 1.0602,
      "step": 8553
    },
    {
      "epoch": 0.6165933828299575,
      "grad_norm": 0.11104689725047194,
      "learning_rate": 6.769397444239416e-05,
      "loss": 1.0657,
      "step": 8554
    },
    {
      "epoch": 0.6166654652922944,
      "grad_norm": 0.12474129884538585,
      "learning_rate": 6.767188008362318e-05,
      "loss": 0.9785,
      "step": 8555
    },
    {
      "epoch": 0.6167375477546313,
      "grad_norm": 0.11191142877440083,
      "learning_rate": 6.764978748702393e-05,
      "loss": 0.9642,
      "step": 8556
    },
    {
      "epoch": 0.6168096302169682,
      "grad_norm": 0.11251191683183649,
      "learning_rate": 6.762769665380067e-05,
      "loss": 1.0824,
      "step": 8557
    },
    {
      "epoch": 0.6168817126793051,
      "grad_norm": 0.1128014094036351,
      "learning_rate": 6.760560758515756e-05,
      "loss": 1.0134,
      "step": 8558
    },
    {
      "epoch": 0.6169537951416421,
      "grad_norm": 0.10755739158873696,
      "learning_rate": 6.758352028229867e-05,
      "loss": 1.0123,
      "step": 8559
    },
    {
      "epoch": 0.617025877603979,
      "grad_norm": 0.13311869515718255,
      "learning_rate": 6.75614347464279e-05,
      "loss": 1.0234,
      "step": 8560
    },
    {
      "epoch": 0.6170979600663159,
      "grad_norm": 0.1312199700253394,
      "learning_rate": 6.753935097874913e-05,
      "loss": 1.0121,
      "step": 8561
    },
    {
      "epoch": 0.6171700425286528,
      "grad_norm": 0.13009000238391252,
      "learning_rate": 6.751726898046617e-05,
      "loss": 0.9606,
      "step": 8562
    },
    {
      "epoch": 0.6172421249909897,
      "grad_norm": 0.15085499578819606,
      "learning_rate": 6.749518875278264e-05,
      "loss": 0.9663,
      "step": 8563
    },
    {
      "epoch": 0.6173142074533267,
      "grad_norm": 0.19399371300648766,
      "learning_rate": 6.747311029690211e-05,
      "loss": 1.0222,
      "step": 8564
    },
    {
      "epoch": 0.6173862899156635,
      "grad_norm": 0.11613119191738049,
      "learning_rate": 6.74510336140281e-05,
      "loss": 0.9702,
      "step": 8565
    },
    {
      "epoch": 0.6174583723780004,
      "grad_norm": 0.12521550773075318,
      "learning_rate": 6.742895870536388e-05,
      "loss": 1.133,
      "step": 8566
    },
    {
      "epoch": 0.6175304548403373,
      "grad_norm": 0.6608236291241821,
      "learning_rate": 6.740688557211284e-05,
      "loss": 1.0508,
      "step": 8567
    },
    {
      "epoch": 0.6176025373026742,
      "grad_norm": 0.1375298141885456,
      "learning_rate": 6.738481421547812e-05,
      "loss": 1.0681,
      "step": 8568
    },
    {
      "epoch": 0.6176746197650111,
      "grad_norm": 0.15000219497400422,
      "learning_rate": 6.736274463666283e-05,
      "loss": 1.0959,
      "step": 8569
    },
    {
      "epoch": 0.6177467022273481,
      "grad_norm": 0.12256033984862015,
      "learning_rate": 6.734067683686986e-05,
      "loss": 1.0528,
      "step": 8570
    },
    {
      "epoch": 0.617818784689685,
      "grad_norm": 0.12773443821942146,
      "learning_rate": 6.731861081730226e-05,
      "loss": 1.0578,
      "step": 8571
    },
    {
      "epoch": 0.6178908671520219,
      "grad_norm": 0.17187326907054862,
      "learning_rate": 6.729654657916272e-05,
      "loss": 1.1481,
      "step": 8572
    },
    {
      "epoch": 0.6179629496143588,
      "grad_norm": 0.1072033943411354,
      "learning_rate": 6.727448412365399e-05,
      "loss": 1.0371,
      "step": 8573
    },
    {
      "epoch": 0.6180350320766957,
      "grad_norm": 0.1233215489445124,
      "learning_rate": 6.725242345197859e-05,
      "loss": 1.0348,
      "step": 8574
    },
    {
      "epoch": 0.6181071145390327,
      "grad_norm": 0.1182672451646766,
      "learning_rate": 6.723036456533913e-05,
      "loss": 1.0531,
      "step": 8575
    },
    {
      "epoch": 0.6181791970013696,
      "grad_norm": 0.10613191921271793,
      "learning_rate": 6.720830746493795e-05,
      "loss": 0.9965,
      "step": 8576
    },
    {
      "epoch": 0.6182512794637065,
      "grad_norm": 0.12606241838345783,
      "learning_rate": 6.718625215197739e-05,
      "loss": 0.9746,
      "step": 8577
    },
    {
      "epoch": 0.6183233619260434,
      "grad_norm": 0.12568394854911025,
      "learning_rate": 6.716419862765963e-05,
      "loss": 1.1088,
      "step": 8578
    },
    {
      "epoch": 0.6183954443883803,
      "grad_norm": 0.22443323086552716,
      "learning_rate": 6.714214689318678e-05,
      "loss": 1.0757,
      "step": 8579
    },
    {
      "epoch": 0.6184675268507173,
      "grad_norm": 0.10904358683425884,
      "learning_rate": 6.712009694976092e-05,
      "loss": 1.0458,
      "step": 8580
    },
    {
      "epoch": 0.6185396093130542,
      "grad_norm": 0.12292533193759456,
      "learning_rate": 6.709804879858392e-05,
      "loss": 0.9933,
      "step": 8581
    },
    {
      "epoch": 0.618611691775391,
      "grad_norm": 0.11429731205774901,
      "learning_rate": 6.707600244085761e-05,
      "loss": 1.0796,
      "step": 8582
    },
    {
      "epoch": 0.6186837742377279,
      "grad_norm": 0.12055842635683629,
      "learning_rate": 6.705395787778368e-05,
      "loss": 1.1017,
      "step": 8583
    },
    {
      "epoch": 0.6187558567000648,
      "grad_norm": 0.11219783536751399,
      "learning_rate": 6.703191511056381e-05,
      "loss": 1.0462,
      "step": 8584
    },
    {
      "epoch": 0.6188279391624018,
      "grad_norm": 0.20437358725266502,
      "learning_rate": 6.700987414039955e-05,
      "loss": 1.1053,
      "step": 8585
    },
    {
      "epoch": 0.6189000216247387,
      "grad_norm": 0.20009609819019195,
      "learning_rate": 6.698783496849224e-05,
      "loss": 1.0761,
      "step": 8586
    },
    {
      "epoch": 0.6189721040870756,
      "grad_norm": 0.11566205250783111,
      "learning_rate": 6.696579759604327e-05,
      "loss": 1.011,
      "step": 8587
    },
    {
      "epoch": 0.6190441865494125,
      "grad_norm": 0.13127872935326232,
      "learning_rate": 6.694376202425385e-05,
      "loss": 1.0613,
      "step": 8588
    },
    {
      "epoch": 0.6191162690117494,
      "grad_norm": 0.1493764258864374,
      "learning_rate": 6.692172825432516e-05,
      "loss": 1.0756,
      "step": 8589
    },
    {
      "epoch": 0.6191883514740864,
      "grad_norm": 0.116946420867736,
      "learning_rate": 6.689969628745819e-05,
      "loss": 1.0692,
      "step": 8590
    },
    {
      "epoch": 0.6192604339364233,
      "grad_norm": 0.10817115749308782,
      "learning_rate": 6.687766612485389e-05,
      "loss": 1.0107,
      "step": 8591
    },
    {
      "epoch": 0.6193325163987602,
      "grad_norm": 0.11492076552929661,
      "learning_rate": 6.68556377677131e-05,
      "loss": 1.039,
      "step": 8592
    },
    {
      "epoch": 0.6194045988610971,
      "grad_norm": 0.12485733598194201,
      "learning_rate": 6.68336112172366e-05,
      "loss": 1.1601,
      "step": 8593
    },
    {
      "epoch": 0.619476681323434,
      "grad_norm": 0.1357793492208886,
      "learning_rate": 6.681158647462499e-05,
      "loss": 0.9444,
      "step": 8594
    },
    {
      "epoch": 0.619548763785771,
      "grad_norm": 0.13268592081556477,
      "learning_rate": 6.678956354107882e-05,
      "loss": 1.0416,
      "step": 8595
    },
    {
      "epoch": 0.6196208462481079,
      "grad_norm": 0.12756699835904467,
      "learning_rate": 6.676754241779852e-05,
      "loss": 1.03,
      "step": 8596
    },
    {
      "epoch": 0.6196929287104448,
      "grad_norm": 0.135370519604329,
      "learning_rate": 6.674552310598453e-05,
      "loss": 1.1024,
      "step": 8597
    },
    {
      "epoch": 0.6197650111727817,
      "grad_norm": 0.10974033010253728,
      "learning_rate": 6.6723505606837e-05,
      "loss": 1.0837,
      "step": 8598
    },
    {
      "epoch": 0.6198370936351186,
      "grad_norm": 0.10943110824083355,
      "learning_rate": 6.670148992155612e-05,
      "loss": 1.0168,
      "step": 8599
    },
    {
      "epoch": 0.6199091760974554,
      "grad_norm": 0.14260629582638726,
      "learning_rate": 6.667947605134193e-05,
      "loss": 0.9916,
      "step": 8600
    },
    {
      "epoch": 0.6199812585597924,
      "grad_norm": 0.17064304294973717,
      "learning_rate": 6.665746399739438e-05,
      "loss": 0.9725,
      "step": 8601
    },
    {
      "epoch": 0.6200533410221293,
      "grad_norm": 0.11838305257883702,
      "learning_rate": 6.663545376091334e-05,
      "loss": 1.0313,
      "step": 8602
    },
    {
      "epoch": 0.6201254234844662,
      "grad_norm": 0.1321300028038331,
      "learning_rate": 6.661344534309856e-05,
      "loss": 1.1127,
      "step": 8603
    },
    {
      "epoch": 0.6201975059468031,
      "grad_norm": 0.11024719342073354,
      "learning_rate": 6.659143874514971e-05,
      "loss": 1.099,
      "step": 8604
    },
    {
      "epoch": 0.62026958840914,
      "grad_norm": 0.12878571176648781,
      "learning_rate": 6.656943396826629e-05,
      "loss": 1.0315,
      "step": 8605
    },
    {
      "epoch": 0.620341670871477,
      "grad_norm": 0.16283774671185847,
      "learning_rate": 6.654743101364783e-05,
      "loss": 1.0548,
      "step": 8606
    },
    {
      "epoch": 0.6204137533338139,
      "grad_norm": 0.15731654230018874,
      "learning_rate": 6.652542988249363e-05,
      "loss": 1.0951,
      "step": 8607
    },
    {
      "epoch": 0.6204858357961508,
      "grad_norm": 0.12608873449391403,
      "learning_rate": 6.650343057600302e-05,
      "loss": 1.032,
      "step": 8608
    },
    {
      "epoch": 0.6205579182584877,
      "grad_norm": 0.12424538259321906,
      "learning_rate": 6.648143309537508e-05,
      "loss": 1.0855,
      "step": 8609
    },
    {
      "epoch": 0.6206300007208246,
      "grad_norm": 0.12655831925079616,
      "learning_rate": 6.645943744180889e-05,
      "loss": 1.0313,
      "step": 8610
    },
    {
      "epoch": 0.6207020831831616,
      "grad_norm": 0.1103907852030594,
      "learning_rate": 6.643744361650342e-05,
      "loss": 1.0902,
      "step": 8611
    },
    {
      "epoch": 0.6207741656454985,
      "grad_norm": 0.11027359194814533,
      "learning_rate": 6.641545162065757e-05,
      "loss": 1.1035,
      "step": 8612
    },
    {
      "epoch": 0.6208462481078354,
      "grad_norm": 0.1321646628381215,
      "learning_rate": 6.639346145547005e-05,
      "loss": 0.9499,
      "step": 8613
    },
    {
      "epoch": 0.6209183305701723,
      "grad_norm": 0.11495614375489459,
      "learning_rate": 6.63714731221395e-05,
      "loss": 1.0314,
      "step": 8614
    },
    {
      "epoch": 0.6209904130325092,
      "grad_norm": 0.12171580886044807,
      "learning_rate": 6.634948662186453e-05,
      "loss": 0.9834,
      "step": 8615
    },
    {
      "epoch": 0.6210624954948462,
      "grad_norm": 0.2111087425908955,
      "learning_rate": 6.632750195584362e-05,
      "loss": 1.1145,
      "step": 8616
    },
    {
      "epoch": 0.621134577957183,
      "grad_norm": 0.12457552050922352,
      "learning_rate": 6.630551912527507e-05,
      "loss": 1.0756,
      "step": 8617
    },
    {
      "epoch": 0.6212066604195199,
      "grad_norm": 0.12516163708224848,
      "learning_rate": 6.628353813135714e-05,
      "loss": 1.0354,
      "step": 8618
    },
    {
      "epoch": 0.6212787428818568,
      "grad_norm": 0.12428348724063637,
      "learning_rate": 6.626155897528805e-05,
      "loss": 1.0199,
      "step": 8619
    },
    {
      "epoch": 0.6213508253441937,
      "grad_norm": 0.12538035515420562,
      "learning_rate": 6.623958165826584e-05,
      "loss": 1.0091,
      "step": 8620
    },
    {
      "epoch": 0.6214229078065306,
      "grad_norm": 0.12014669211678453,
      "learning_rate": 6.621760618148845e-05,
      "loss": 1.038,
      "step": 8621
    },
    {
      "epoch": 0.6214949902688676,
      "grad_norm": 0.11804931848104676,
      "learning_rate": 6.619563254615375e-05,
      "loss": 1.0397,
      "step": 8622
    },
    {
      "epoch": 0.6215670727312045,
      "grad_norm": 0.11502469425498012,
      "learning_rate": 6.617366075345944e-05,
      "loss": 1.056,
      "step": 8623
    },
    {
      "epoch": 0.6216391551935414,
      "grad_norm": 0.12978740339568492,
      "learning_rate": 6.615169080460331e-05,
      "loss": 1.0286,
      "step": 8624
    },
    {
      "epoch": 0.6217112376558783,
      "grad_norm": 0.11935463647662259,
      "learning_rate": 6.612972270078282e-05,
      "loss": 1.0624,
      "step": 8625
    },
    {
      "epoch": 0.6217833201182152,
      "grad_norm": 0.11923589158400894,
      "learning_rate": 6.610775644319549e-05,
      "loss": 1.0892,
      "step": 8626
    },
    {
      "epoch": 0.6218554025805522,
      "grad_norm": 0.12308436145645023,
      "learning_rate": 6.608579203303858e-05,
      "loss": 0.9948,
      "step": 8627
    },
    {
      "epoch": 0.6219274850428891,
      "grad_norm": 0.19150005335248313,
      "learning_rate": 6.606382947150944e-05,
      "loss": 1.0222,
      "step": 8628
    },
    {
      "epoch": 0.621999567505226,
      "grad_norm": 0.11350403750908375,
      "learning_rate": 6.60418687598052e-05,
      "loss": 1.0558,
      "step": 8629
    },
    {
      "epoch": 0.6220716499675629,
      "grad_norm": 0.11717748174196754,
      "learning_rate": 6.601990989912293e-05,
      "loss": 1.0961,
      "step": 8630
    },
    {
      "epoch": 0.6221437324298998,
      "grad_norm": 0.12309562977359019,
      "learning_rate": 6.599795289065954e-05,
      "loss": 1.1175,
      "step": 8631
    },
    {
      "epoch": 0.6222158148922368,
      "grad_norm": 0.17634059243807798,
      "learning_rate": 6.59759977356119e-05,
      "loss": 1.0305,
      "step": 8632
    },
    {
      "epoch": 0.6222878973545737,
      "grad_norm": 0.12531254648128526,
      "learning_rate": 6.595404443517678e-05,
      "loss": 0.9642,
      "step": 8633
    },
    {
      "epoch": 0.6223599798169105,
      "grad_norm": 0.12128828363746345,
      "learning_rate": 6.593209299055083e-05,
      "loss": 1.0122,
      "step": 8634
    },
    {
      "epoch": 0.6224320622792474,
      "grad_norm": 0.1295736180651181,
      "learning_rate": 6.591014340293059e-05,
      "loss": 1.081,
      "step": 8635
    },
    {
      "epoch": 0.6225041447415843,
      "grad_norm": 0.12822057774363602,
      "learning_rate": 6.588819567351248e-05,
      "loss": 1.0066,
      "step": 8636
    },
    {
      "epoch": 0.6225762272039213,
      "grad_norm": 0.12442262219992987,
      "learning_rate": 6.586624980349292e-05,
      "loss": 1.086,
      "step": 8637
    },
    {
      "epoch": 0.6226483096662582,
      "grad_norm": 0.39007341655902067,
      "learning_rate": 6.584430579406812e-05,
      "loss": 1.0766,
      "step": 8638
    },
    {
      "epoch": 0.6227203921285951,
      "grad_norm": 0.16808815049734985,
      "learning_rate": 6.58223636464342e-05,
      "loss": 0.969,
      "step": 8639
    },
    {
      "epoch": 0.622792474590932,
      "grad_norm": 0.13884027931356274,
      "learning_rate": 6.58004233617872e-05,
      "loss": 0.9945,
      "step": 8640
    },
    {
      "epoch": 0.6228645570532689,
      "grad_norm": 0.1179124284724607,
      "learning_rate": 6.57784849413231e-05,
      "loss": 1.1011,
      "step": 8641
    },
    {
      "epoch": 0.6229366395156059,
      "grad_norm": 0.10098437749176806,
      "learning_rate": 6.575654838623776e-05,
      "loss": 0.9932,
      "step": 8642
    },
    {
      "epoch": 0.6230087219779428,
      "grad_norm": 0.1828144130590869,
      "learning_rate": 6.573461369772685e-05,
      "loss": 1.031,
      "step": 8643
    },
    {
      "epoch": 0.6230808044402797,
      "grad_norm": 0.14302402852541507,
      "learning_rate": 6.571268087698608e-05,
      "loss": 1.116,
      "step": 8644
    },
    {
      "epoch": 0.6231528869026166,
      "grad_norm": 0.18141885069984645,
      "learning_rate": 6.569074992521087e-05,
      "loss": 1.0926,
      "step": 8645
    },
    {
      "epoch": 0.6232249693649535,
      "grad_norm": 0.1351921259171278,
      "learning_rate": 6.566882084359678e-05,
      "loss": 1.0383,
      "step": 8646
    },
    {
      "epoch": 0.6232970518272904,
      "grad_norm": 0.12979289137365202,
      "learning_rate": 6.56468936333391e-05,
      "loss": 1.0776,
      "step": 8647
    },
    {
      "epoch": 0.6233691342896274,
      "grad_norm": 0.11774293845479467,
      "learning_rate": 6.562496829563304e-05,
      "loss": 1.1223,
      "step": 8648
    },
    {
      "epoch": 0.6234412167519643,
      "grad_norm": 0.11122523888553987,
      "learning_rate": 6.560304483167368e-05,
      "loss": 1.0701,
      "step": 8649
    },
    {
      "epoch": 0.6235132992143012,
      "grad_norm": 0.1084710170205144,
      "learning_rate": 6.558112324265617e-05,
      "loss": 1.0403,
      "step": 8650
    },
    {
      "epoch": 0.6235853816766381,
      "grad_norm": 0.1288640199824428,
      "learning_rate": 6.555920352977534e-05,
      "loss": 1.1089,
      "step": 8651
    },
    {
      "epoch": 0.6236574641389749,
      "grad_norm": 0.11945960695472171,
      "learning_rate": 6.553728569422604e-05,
      "loss": 1.024,
      "step": 8652
    },
    {
      "epoch": 0.6237295466013119,
      "grad_norm": 0.13427719053676754,
      "learning_rate": 6.551536973720298e-05,
      "loss": 1.0004,
      "step": 8653
    },
    {
      "epoch": 0.6238016290636488,
      "grad_norm": 0.12497122269947601,
      "learning_rate": 6.549345565990075e-05,
      "loss": 1.05,
      "step": 8654
    },
    {
      "epoch": 0.6238737115259857,
      "grad_norm": 0.14226982876622984,
      "learning_rate": 6.547154346351392e-05,
      "loss": 1.0311,
      "step": 8655
    },
    {
      "epoch": 0.6239457939883226,
      "grad_norm": 0.12203460272163175,
      "learning_rate": 6.544963314923689e-05,
      "loss": 1.0451,
      "step": 8656
    },
    {
      "epoch": 0.6240178764506595,
      "grad_norm": 0.20438468838053786,
      "learning_rate": 6.542772471826391e-05,
      "loss": 1.0198,
      "step": 8657
    },
    {
      "epoch": 0.6240899589129965,
      "grad_norm": 0.14259112982565986,
      "learning_rate": 6.540581817178923e-05,
      "loss": 1.0319,
      "step": 8658
    },
    {
      "epoch": 0.6241620413753334,
      "grad_norm": 0.3200958625276833,
      "learning_rate": 6.538391351100696e-05,
      "loss": 1.0031,
      "step": 8659
    },
    {
      "epoch": 0.6242341238376703,
      "grad_norm": 0.10784066737682084,
      "learning_rate": 6.53620107371111e-05,
      "loss": 1.0649,
      "step": 8660
    },
    {
      "epoch": 0.6243062063000072,
      "grad_norm": 0.16036835060678406,
      "learning_rate": 6.534010985129552e-05,
      "loss": 0.9476,
      "step": 8661
    },
    {
      "epoch": 0.6243782887623441,
      "grad_norm": 0.12600169974730058,
      "learning_rate": 6.531821085475401e-05,
      "loss": 1.0529,
      "step": 8662
    },
    {
      "epoch": 0.6244503712246811,
      "grad_norm": 0.11890234872600904,
      "learning_rate": 6.529631374868031e-05,
      "loss": 1.0509,
      "step": 8663
    },
    {
      "epoch": 0.624522453687018,
      "grad_norm": 0.12877791137447095,
      "learning_rate": 6.527441853426799e-05,
      "loss": 1.0846,
      "step": 8664
    },
    {
      "epoch": 0.6245945361493549,
      "grad_norm": 0.14838132140635685,
      "learning_rate": 6.52525252127105e-05,
      "loss": 1.0635,
      "step": 8665
    },
    {
      "epoch": 0.6246666186116918,
      "grad_norm": 0.11201783847146984,
      "learning_rate": 6.523063378520124e-05,
      "loss": 1.1034,
      "step": 8666
    },
    {
      "epoch": 0.6247387010740287,
      "grad_norm": 0.10787629364938171,
      "learning_rate": 6.520874425293348e-05,
      "loss": 1.0409,
      "step": 8667
    },
    {
      "epoch": 0.6248107835363657,
      "grad_norm": 0.12114654082832073,
      "learning_rate": 6.518685661710046e-05,
      "loss": 1.0411,
      "step": 8668
    },
    {
      "epoch": 0.6248828659987025,
      "grad_norm": 0.11213880191624946,
      "learning_rate": 6.516497087889517e-05,
      "loss": 1.0863,
      "step": 8669
    },
    {
      "epoch": 0.6249549484610394,
      "grad_norm": 0.12589354432356162,
      "learning_rate": 6.514308703951062e-05,
      "loss": 1.0646,
      "step": 8670
    },
    {
      "epoch": 0.6250270309233763,
      "grad_norm": 0.12037943286864076,
      "learning_rate": 6.512120510013963e-05,
      "loss": 1.0936,
      "step": 8671
    },
    {
      "epoch": 0.6250991133857132,
      "grad_norm": 0.13332399945694706,
      "learning_rate": 6.509932506197505e-05,
      "loss": 1.0492,
      "step": 8672
    },
    {
      "epoch": 0.6251711958480501,
      "grad_norm": 0.11910088290270035,
      "learning_rate": 6.507744692620947e-05,
      "loss": 0.9152,
      "step": 8673
    },
    {
      "epoch": 0.6252432783103871,
      "grad_norm": 0.12550327429318867,
      "learning_rate": 6.505557069403545e-05,
      "loss": 1.0895,
      "step": 8674
    },
    {
      "epoch": 0.625315360772724,
      "grad_norm": 0.14248929131874244,
      "learning_rate": 6.503369636664547e-05,
      "loss": 1.1127,
      "step": 8675
    },
    {
      "epoch": 0.6253874432350609,
      "grad_norm": 0.1311541466268362,
      "learning_rate": 6.501182394523182e-05,
      "loss": 1.0345,
      "step": 8676
    },
    {
      "epoch": 0.6254595256973978,
      "grad_norm": 0.21903438498787062,
      "learning_rate": 6.498995343098679e-05,
      "loss": 1.1159,
      "step": 8677
    },
    {
      "epoch": 0.6255316081597347,
      "grad_norm": 0.12698609513129325,
      "learning_rate": 6.49680848251025e-05,
      "loss": 1.0074,
      "step": 8678
    },
    {
      "epoch": 0.6256036906220717,
      "grad_norm": 0.12929690838736993,
      "learning_rate": 6.494621812877102e-05,
      "loss": 1.1045,
      "step": 8679
    },
    {
      "epoch": 0.6256757730844086,
      "grad_norm": 0.114921975436813,
      "learning_rate": 6.492435334318422e-05,
      "loss": 1.0496,
      "step": 8680
    },
    {
      "epoch": 0.6257478555467455,
      "grad_norm": 0.10949760141493915,
      "learning_rate": 6.490249046953397e-05,
      "loss": 1.0622,
      "step": 8681
    },
    {
      "epoch": 0.6258199380090824,
      "grad_norm": 0.2203611223289161,
      "learning_rate": 6.488062950901198e-05,
      "loss": 1.0411,
      "step": 8682
    },
    {
      "epoch": 0.6258920204714193,
      "grad_norm": 0.12175476745130813,
      "learning_rate": 6.485877046280989e-05,
      "loss": 1.0219,
      "step": 8683
    },
    {
      "epoch": 0.6259641029337563,
      "grad_norm": 0.11610430700938101,
      "learning_rate": 6.483691333211915e-05,
      "loss": 1.0209,
      "step": 8684
    },
    {
      "epoch": 0.6260361853960932,
      "grad_norm": 0.1518838196136037,
      "learning_rate": 6.481505811813125e-05,
      "loss": 1.1338,
      "step": 8685
    },
    {
      "epoch": 0.62610826785843,
      "grad_norm": 0.11372685459591651,
      "learning_rate": 6.479320482203744e-05,
      "loss": 1.0574,
      "step": 8686
    },
    {
      "epoch": 0.6261803503207669,
      "grad_norm": 0.1278352354737097,
      "learning_rate": 6.477135344502895e-05,
      "loss": 1.0056,
      "step": 8687
    },
    {
      "epoch": 0.6262524327831038,
      "grad_norm": 0.44265800073068096,
      "learning_rate": 6.474950398829684e-05,
      "loss": 1.0882,
      "step": 8688
    },
    {
      "epoch": 0.6263245152454407,
      "grad_norm": 0.1364222489386907,
      "learning_rate": 6.472765645303212e-05,
      "loss": 1.0675,
      "step": 8689
    },
    {
      "epoch": 0.6263965977077777,
      "grad_norm": 0.11180337808929085,
      "learning_rate": 6.470581084042568e-05,
      "loss": 1.0603,
      "step": 8690
    },
    {
      "epoch": 0.6264686801701146,
      "grad_norm": 0.12340349896989661,
      "learning_rate": 6.468396715166834e-05,
      "loss": 1.0562,
      "step": 8691
    },
    {
      "epoch": 0.6265407626324515,
      "grad_norm": 0.13148808831685826,
      "learning_rate": 6.46621253879507e-05,
      "loss": 1.0786,
      "step": 8692
    },
    {
      "epoch": 0.6266128450947884,
      "grad_norm": 0.12524043123321393,
      "learning_rate": 6.464028555046335e-05,
      "loss": 1.0063,
      "step": 8693
    },
    {
      "epoch": 0.6266849275571253,
      "grad_norm": 0.11833638087145094,
      "learning_rate": 6.46184476403968e-05,
      "loss": 0.9902,
      "step": 8694
    },
    {
      "epoch": 0.6267570100194623,
      "grad_norm": 0.11126022104058018,
      "learning_rate": 6.459661165894141e-05,
      "loss": 1.0167,
      "step": 8695
    },
    {
      "epoch": 0.6268290924817992,
      "grad_norm": 0.12668395093970347,
      "learning_rate": 6.457477760728738e-05,
      "loss": 1.0901,
      "step": 8696
    },
    {
      "epoch": 0.6269011749441361,
      "grad_norm": 0.12049238150683593,
      "learning_rate": 6.455294548662492e-05,
      "loss": 0.9335,
      "step": 8697
    },
    {
      "epoch": 0.626973257406473,
      "grad_norm": 0.1080449647541313,
      "learning_rate": 6.4531115298144e-05,
      "loss": 1.0433,
      "step": 8698
    },
    {
      "epoch": 0.62704533986881,
      "grad_norm": 0.12544175411907663,
      "learning_rate": 6.450928704303465e-05,
      "loss": 1.0213,
      "step": 8699
    },
    {
      "epoch": 0.6271174223311469,
      "grad_norm": 0.1794176103340038,
      "learning_rate": 6.448746072248666e-05,
      "loss": 1.0506,
      "step": 8700
    },
    {
      "epoch": 0.6271895047934838,
      "grad_norm": 0.13424135558053876,
      "learning_rate": 6.446563633768979e-05,
      "loss": 1.0412,
      "step": 8701
    },
    {
      "epoch": 0.6272615872558207,
      "grad_norm": 0.12827021337766678,
      "learning_rate": 6.444381388983356e-05,
      "loss": 1.0004,
      "step": 8702
    },
    {
      "epoch": 0.6273336697181576,
      "grad_norm": 0.12383726333252992,
      "learning_rate": 6.442199338010764e-05,
      "loss": 1.0562,
      "step": 8703
    },
    {
      "epoch": 0.6274057521804944,
      "grad_norm": 0.11078533271600692,
      "learning_rate": 6.440017480970135e-05,
      "loss": 1.03,
      "step": 8704
    },
    {
      "epoch": 0.6274778346428314,
      "grad_norm": 0.10902585296621935,
      "learning_rate": 6.437835817980403e-05,
      "loss": 1.0389,
      "step": 8705
    },
    {
      "epoch": 0.6275499171051683,
      "grad_norm": 0.11039072398799266,
      "learning_rate": 6.435654349160482e-05,
      "loss": 1.0727,
      "step": 8706
    },
    {
      "epoch": 0.6276219995675052,
      "grad_norm": 1.5912961524298994,
      "learning_rate": 6.433473074629291e-05,
      "loss": 1.0332,
      "step": 8707
    },
    {
      "epoch": 0.6276940820298421,
      "grad_norm": 0.12631385290107383,
      "learning_rate": 6.431291994505724e-05,
      "loss": 1.1182,
      "step": 8708
    },
    {
      "epoch": 0.627766164492179,
      "grad_norm": 0.1249767613353299,
      "learning_rate": 6.429111108908673e-05,
      "loss": 1.1134,
      "step": 8709
    },
    {
      "epoch": 0.627838246954516,
      "grad_norm": 0.12543481006388227,
      "learning_rate": 6.426930417957007e-05,
      "loss": 1.0786,
      "step": 8710
    },
    {
      "epoch": 0.6279103294168529,
      "grad_norm": 0.12998327225370374,
      "learning_rate": 6.4247499217696e-05,
      "loss": 1.0602,
      "step": 8711
    },
    {
      "epoch": 0.6279824118791898,
      "grad_norm": 0.10411697191959948,
      "learning_rate": 6.422569620465308e-05,
      "loss": 1.0422,
      "step": 8712
    },
    {
      "epoch": 0.6280544943415267,
      "grad_norm": 0.14638433183060506,
      "learning_rate": 6.420389514162979e-05,
      "loss": 1.1535,
      "step": 8713
    },
    {
      "epoch": 0.6281265768038636,
      "grad_norm": 0.1444471343767057,
      "learning_rate": 6.418209602981442e-05,
      "loss": 1.1658,
      "step": 8714
    },
    {
      "epoch": 0.6281986592662006,
      "grad_norm": 0.1189922731798444,
      "learning_rate": 6.416029887039525e-05,
      "loss": 0.9998,
      "step": 8715
    },
    {
      "epoch": 0.6282707417285375,
      "grad_norm": 0.11704041109013899,
      "learning_rate": 6.413850366456043e-05,
      "loss": 1.0213,
      "step": 8716
    },
    {
      "epoch": 0.6283428241908744,
      "grad_norm": 0.178397625087582,
      "learning_rate": 6.411671041349801e-05,
      "loss": 1.0174,
      "step": 8717
    },
    {
      "epoch": 0.6284149066532113,
      "grad_norm": 0.11595105683424797,
      "learning_rate": 6.409491911839588e-05,
      "loss": 1.0754,
      "step": 8718
    },
    {
      "epoch": 0.6284869891155482,
      "grad_norm": 0.11690557647273053,
      "learning_rate": 6.407312978044188e-05,
      "loss": 1.0345,
      "step": 8719
    },
    {
      "epoch": 0.6285590715778852,
      "grad_norm": 0.11709730308533636,
      "learning_rate": 6.405134240082368e-05,
      "loss": 1.0847,
      "step": 8720
    },
    {
      "epoch": 0.628631154040222,
      "grad_norm": 0.11474613237820751,
      "learning_rate": 6.402955698072896e-05,
      "loss": 1.0395,
      "step": 8721
    },
    {
      "epoch": 0.6287032365025589,
      "grad_norm": 0.14712960413746584,
      "learning_rate": 6.400777352134519e-05,
      "loss": 0.9289,
      "step": 8722
    },
    {
      "epoch": 0.6287753189648958,
      "grad_norm": 0.12761638698778852,
      "learning_rate": 6.398599202385978e-05,
      "loss": 1.0437,
      "step": 8723
    },
    {
      "epoch": 0.6288474014272327,
      "grad_norm": 0.12389133409455164,
      "learning_rate": 6.39642124894599e-05,
      "loss": 0.9701,
      "step": 8724
    },
    {
      "epoch": 0.6289194838895696,
      "grad_norm": 0.1805198304061247,
      "learning_rate": 6.394243491933292e-05,
      "loss": 1.0361,
      "step": 8725
    },
    {
      "epoch": 0.6289915663519066,
      "grad_norm": 0.11230222528130439,
      "learning_rate": 6.392065931466578e-05,
      "loss": 1.0246,
      "step": 8726
    },
    {
      "epoch": 0.6290636488142435,
      "grad_norm": 0.4798976632971514,
      "learning_rate": 6.38988856766455e-05,
      "loss": 1.0408,
      "step": 8727
    },
    {
      "epoch": 0.6291357312765804,
      "grad_norm": 0.12531724331619754,
      "learning_rate": 6.387711400645888e-05,
      "loss": 1.0754,
      "step": 8728
    },
    {
      "epoch": 0.6292078137389173,
      "grad_norm": 0.1677602935026547,
      "learning_rate": 6.385534430529275e-05,
      "loss": 1.0421,
      "step": 8729
    },
    {
      "epoch": 0.6292798962012542,
      "grad_norm": 0.11034714735329317,
      "learning_rate": 6.383357657433371e-05,
      "loss": 1.067,
      "step": 8730
    },
    {
      "epoch": 0.6293519786635912,
      "grad_norm": 0.14043235188937472,
      "learning_rate": 6.38118108147683e-05,
      "loss": 1.0143,
      "step": 8731
    },
    {
      "epoch": 0.6294240611259281,
      "grad_norm": 0.10987733400253552,
      "learning_rate": 6.379004702778295e-05,
      "loss": 1.03,
      "step": 8732
    },
    {
      "epoch": 0.629496143588265,
      "grad_norm": 0.11632504319865943,
      "learning_rate": 6.376828521456395e-05,
      "loss": 1.0732,
      "step": 8733
    },
    {
      "epoch": 0.6295682260506019,
      "grad_norm": 0.11679425083373188,
      "learning_rate": 6.374652537629757e-05,
      "loss": 0.999,
      "step": 8734
    },
    {
      "epoch": 0.6296403085129388,
      "grad_norm": 0.12087859961890408,
      "learning_rate": 6.372476751416991e-05,
      "loss": 1.0238,
      "step": 8735
    },
    {
      "epoch": 0.6297123909752758,
      "grad_norm": 0.11400351706991112,
      "learning_rate": 6.370301162936693e-05,
      "loss": 1.0284,
      "step": 8736
    },
    {
      "epoch": 0.6297844734376127,
      "grad_norm": 0.1187557080443888,
      "learning_rate": 6.368125772307452e-05,
      "loss": 1.0396,
      "step": 8737
    },
    {
      "epoch": 0.6298565558999495,
      "grad_norm": 0.17540955025931312,
      "learning_rate": 6.365950579647848e-05,
      "loss": 1.0695,
      "step": 8738
    },
    {
      "epoch": 0.6299286383622864,
      "grad_norm": 0.12211401072341235,
      "learning_rate": 6.363775585076452e-05,
      "loss": 0.9967,
      "step": 8739
    },
    {
      "epoch": 0.6300007208246233,
      "grad_norm": 0.1255592675251765,
      "learning_rate": 6.361600788711816e-05,
      "loss": 1.1261,
      "step": 8740
    },
    {
      "epoch": 0.6300728032869602,
      "grad_norm": 0.11896886058758238,
      "learning_rate": 6.359426190672487e-05,
      "loss": 1.0642,
      "step": 8741
    },
    {
      "epoch": 0.6301448857492972,
      "grad_norm": 0.13302975193503713,
      "learning_rate": 6.357251791076997e-05,
      "loss": 1.0269,
      "step": 8742
    },
    {
      "epoch": 0.6302169682116341,
      "grad_norm": 0.10101230401690717,
      "learning_rate": 6.355077590043876e-05,
      "loss": 0.9966,
      "step": 8743
    },
    {
      "epoch": 0.630289050673971,
      "grad_norm": 0.1285030671085674,
      "learning_rate": 6.352903587691635e-05,
      "loss": 1.0294,
      "step": 8744
    },
    {
      "epoch": 0.6303611331363079,
      "grad_norm": 0.12617191367604322,
      "learning_rate": 6.350729784138776e-05,
      "loss": 1.0633,
      "step": 8745
    },
    {
      "epoch": 0.6304332155986448,
      "grad_norm": 0.12471942844398702,
      "learning_rate": 6.348556179503788e-05,
      "loss": 1.067,
      "step": 8746
    },
    {
      "epoch": 0.6305052980609818,
      "grad_norm": 0.11284410816046038,
      "learning_rate": 6.346382773905162e-05,
      "loss": 0.987,
      "step": 8747
    },
    {
      "epoch": 0.6305773805233187,
      "grad_norm": 0.24908042450573836,
      "learning_rate": 6.344209567461357e-05,
      "loss": 1.066,
      "step": 8748
    },
    {
      "epoch": 0.6306494629856556,
      "grad_norm": 0.1394666614915874,
      "learning_rate": 6.342036560290835e-05,
      "loss": 1.0306,
      "step": 8749
    },
    {
      "epoch": 0.6307215454479925,
      "grad_norm": 0.15096949738037524,
      "learning_rate": 6.339863752512046e-05,
      "loss": 1.082,
      "step": 8750
    },
    {
      "epoch": 0.6307936279103294,
      "grad_norm": 1.8237300196389783,
      "learning_rate": 6.337691144243429e-05,
      "loss": 0.9896,
      "step": 8751
    },
    {
      "epoch": 0.6308657103726664,
      "grad_norm": 0.11365595596962369,
      "learning_rate": 6.335518735603407e-05,
      "loss": 1.0089,
      "step": 8752
    },
    {
      "epoch": 0.6309377928350033,
      "grad_norm": 0.11911754461513292,
      "learning_rate": 6.333346526710398e-05,
      "loss": 1.0792,
      "step": 8753
    },
    {
      "epoch": 0.6310098752973402,
      "grad_norm": 0.1230475081654189,
      "learning_rate": 6.331174517682806e-05,
      "loss": 1.071,
      "step": 8754
    },
    {
      "epoch": 0.6310819577596771,
      "grad_norm": 0.12258674042817519,
      "learning_rate": 6.329002708639022e-05,
      "loss": 1.0301,
      "step": 8755
    },
    {
      "epoch": 0.6311540402220139,
      "grad_norm": 0.13158284298494147,
      "learning_rate": 6.326831099697435e-05,
      "loss": 0.972,
      "step": 8756
    },
    {
      "epoch": 0.6312261226843509,
      "grad_norm": 0.5487422821741718,
      "learning_rate": 6.324659690976414e-05,
      "loss": 1.0714,
      "step": 8757
    },
    {
      "epoch": 0.6312982051466878,
      "grad_norm": 0.1226537353570546,
      "learning_rate": 6.322488482594323e-05,
      "loss": 1.0679,
      "step": 8758
    },
    {
      "epoch": 0.6313702876090247,
      "grad_norm": 0.12744993762146903,
      "learning_rate": 6.320317474669504e-05,
      "loss": 1.03,
      "step": 8759
    },
    {
      "epoch": 0.6314423700713616,
      "grad_norm": 0.1244011532818525,
      "learning_rate": 6.318146667320306e-05,
      "loss": 1.0295,
      "step": 8760
    },
    {
      "epoch": 0.6315144525336985,
      "grad_norm": 0.11188901145490078,
      "learning_rate": 6.315976060665052e-05,
      "loss": 1.0744,
      "step": 8761
    },
    {
      "epoch": 0.6315865349960355,
      "grad_norm": 0.16125657957342487,
      "learning_rate": 6.313805654822063e-05,
      "loss": 0.9924,
      "step": 8762
    },
    {
      "epoch": 0.6316586174583724,
      "grad_norm": 0.15455082707849258,
      "learning_rate": 6.311635449909642e-05,
      "loss": 1.0438,
      "step": 8763
    },
    {
      "epoch": 0.6317306999207093,
      "grad_norm": 0.11818080862711662,
      "learning_rate": 6.309465446046083e-05,
      "loss": 1.0579,
      "step": 8764
    },
    {
      "epoch": 0.6318027823830462,
      "grad_norm": 0.12838265849724048,
      "learning_rate": 6.307295643349678e-05,
      "loss": 0.9876,
      "step": 8765
    },
    {
      "epoch": 0.6318748648453831,
      "grad_norm": 0.33391173784662137,
      "learning_rate": 6.305126041938696e-05,
      "loss": 1.0683,
      "step": 8766
    },
    {
      "epoch": 0.63194694730772,
      "grad_norm": 0.16054672174038687,
      "learning_rate": 6.3029566419314e-05,
      "loss": 1.0175,
      "step": 8767
    },
    {
      "epoch": 0.632019029770057,
      "grad_norm": 0.12599238688912998,
      "learning_rate": 6.300787443446036e-05,
      "loss": 1.1432,
      "step": 8768
    },
    {
      "epoch": 0.6320911122323939,
      "grad_norm": 0.11386307030366859,
      "learning_rate": 6.298618446600856e-05,
      "loss": 1.0351,
      "step": 8769
    },
    {
      "epoch": 0.6321631946947308,
      "grad_norm": 0.11695453430222387,
      "learning_rate": 6.296449651514085e-05,
      "loss": 0.995,
      "step": 8770
    },
    {
      "epoch": 0.6322352771570677,
      "grad_norm": 0.14139999573231116,
      "learning_rate": 6.294281058303939e-05,
      "loss": 0.999,
      "step": 8771
    },
    {
      "epoch": 0.6323073596194047,
      "grad_norm": 0.12228752280093705,
      "learning_rate": 6.292112667088625e-05,
      "loss": 1.0442,
      "step": 8772
    },
    {
      "epoch": 0.6323794420817415,
      "grad_norm": 0.11449332620494554,
      "learning_rate": 6.289944477986345e-05,
      "loss": 1.0284,
      "step": 8773
    },
    {
      "epoch": 0.6324515245440784,
      "grad_norm": 0.12047741944176589,
      "learning_rate": 6.287776491115285e-05,
      "loss": 1.1053,
      "step": 8774
    },
    {
      "epoch": 0.6325236070064153,
      "grad_norm": 0.12919639605910896,
      "learning_rate": 6.285608706593614e-05,
      "loss": 1.0835,
      "step": 8775
    },
    {
      "epoch": 0.6325956894687522,
      "grad_norm": 0.14030914842771014,
      "learning_rate": 6.2834411245395e-05,
      "loss": 1.1075,
      "step": 8776
    },
    {
      "epoch": 0.6326677719310891,
      "grad_norm": 0.13757679300080072,
      "learning_rate": 6.281273745071088e-05,
      "loss": 1.0835,
      "step": 8777
    },
    {
      "epoch": 0.6327398543934261,
      "grad_norm": 0.11728658810437649,
      "learning_rate": 6.279106568306531e-05,
      "loss": 0.9875,
      "step": 8778
    },
    {
      "epoch": 0.632811936855763,
      "grad_norm": 0.1363747863195624,
      "learning_rate": 6.276939594363954e-05,
      "loss": 1.063,
      "step": 8779
    },
    {
      "epoch": 0.6328840193180999,
      "grad_norm": 0.12883629843671804,
      "learning_rate": 6.274772823361477e-05,
      "loss": 1.0922,
      "step": 8780
    },
    {
      "epoch": 0.6329561017804368,
      "grad_norm": 0.1168628801844692,
      "learning_rate": 6.272606255417202e-05,
      "loss": 1.0592,
      "step": 8781
    },
    {
      "epoch": 0.6330281842427737,
      "grad_norm": 0.11903449599629977,
      "learning_rate": 6.270439890649238e-05,
      "loss": 1.006,
      "step": 8782
    },
    {
      "epoch": 0.6331002667051107,
      "grad_norm": 0.11636044995285952,
      "learning_rate": 6.268273729175664e-05,
      "loss": 1.0662,
      "step": 8783
    },
    {
      "epoch": 0.6331723491674476,
      "grad_norm": 0.25323721167190394,
      "learning_rate": 6.266107771114559e-05,
      "loss": 1.1203,
      "step": 8784
    },
    {
      "epoch": 0.6332444316297845,
      "grad_norm": 0.11015363605038622,
      "learning_rate": 6.263942016583982e-05,
      "loss": 1.0402,
      "step": 8785
    },
    {
      "epoch": 0.6333165140921214,
      "grad_norm": 0.13936977338559445,
      "learning_rate": 6.261776465701987e-05,
      "loss": 1.003,
      "step": 8786
    },
    {
      "epoch": 0.6333885965544583,
      "grad_norm": 0.14765754644446938,
      "learning_rate": 6.259611118586621e-05,
      "loss": 1.0394,
      "step": 8787
    },
    {
      "epoch": 0.6334606790167953,
      "grad_norm": 0.1447748620148841,
      "learning_rate": 6.257445975355913e-05,
      "loss": 1.0904,
      "step": 8788
    },
    {
      "epoch": 0.6335327614791322,
      "grad_norm": 0.11957259948705613,
      "learning_rate": 6.255281036127879e-05,
      "loss": 1.0369,
      "step": 8789
    },
    {
      "epoch": 0.633604843941469,
      "grad_norm": 0.1498011075265897,
      "learning_rate": 6.253116301020527e-05,
      "loss": 1.0722,
      "step": 8790
    },
    {
      "epoch": 0.6336769264038059,
      "grad_norm": 0.10778006578692476,
      "learning_rate": 6.250951770151861e-05,
      "loss": 1.0513,
      "step": 8791
    },
    {
      "epoch": 0.6337490088661428,
      "grad_norm": 0.12042447932589463,
      "learning_rate": 6.248787443639863e-05,
      "loss": 1.1033,
      "step": 8792
    },
    {
      "epoch": 0.6338210913284797,
      "grad_norm": 0.1295662431929872,
      "learning_rate": 6.24662332160251e-05,
      "loss": 0.9652,
      "step": 8793
    },
    {
      "epoch": 0.6338931737908167,
      "grad_norm": 0.23250506119115186,
      "learning_rate": 6.244459404157758e-05,
      "loss": 1.0353,
      "step": 8794
    },
    {
      "epoch": 0.6339652562531536,
      "grad_norm": 0.1268045690378719,
      "learning_rate": 6.242295691423572e-05,
      "loss": 1.0333,
      "step": 8795
    },
    {
      "epoch": 0.6340373387154905,
      "grad_norm": 0.11284707001571852,
      "learning_rate": 6.24013218351789e-05,
      "loss": 1.0543,
      "step": 8796
    },
    {
      "epoch": 0.6341094211778274,
      "grad_norm": 0.11390340879087382,
      "learning_rate": 6.237968880558637e-05,
      "loss": 0.98,
      "step": 8797
    },
    {
      "epoch": 0.6341815036401643,
      "grad_norm": 0.18659836862911774,
      "learning_rate": 6.23580578266374e-05,
      "loss": 1.0237,
      "step": 8798
    },
    {
      "epoch": 0.6342535861025013,
      "grad_norm": 0.125109564821104,
      "learning_rate": 6.233642889951097e-05,
      "loss": 1.0916,
      "step": 8799
    },
    {
      "epoch": 0.6343256685648382,
      "grad_norm": 0.15384204055936515,
      "learning_rate": 6.231480202538615e-05,
      "loss": 1.001,
      "step": 8800
    },
    {
      "epoch": 0.6343977510271751,
      "grad_norm": 0.1643574412059329,
      "learning_rate": 6.229317720544176e-05,
      "loss": 1.1521,
      "step": 8801
    },
    {
      "epoch": 0.634469833489512,
      "grad_norm": 0.14085160807707528,
      "learning_rate": 6.227155444085658e-05,
      "loss": 1.0855,
      "step": 8802
    },
    {
      "epoch": 0.634541915951849,
      "grad_norm": 0.10475054906849435,
      "learning_rate": 6.224993373280912e-05,
      "loss": 0.9598,
      "step": 8803
    },
    {
      "epoch": 0.6346139984141859,
      "grad_norm": 0.13279148848544275,
      "learning_rate": 6.222831508247808e-05,
      "loss": 0.9589,
      "step": 8804
    },
    {
      "epoch": 0.6346860808765228,
      "grad_norm": 0.11946274845201144,
      "learning_rate": 6.220669849104175e-05,
      "loss": 0.9776,
      "step": 8805
    },
    {
      "epoch": 0.6347581633388597,
      "grad_norm": 0.11720987379132888,
      "learning_rate": 6.218508395967849e-05,
      "loss": 1.0699,
      "step": 8806
    },
    {
      "epoch": 0.6348302458011966,
      "grad_norm": 0.2002893249742608,
      "learning_rate": 6.216347148956643e-05,
      "loss": 1.097,
      "step": 8807
    },
    {
      "epoch": 0.6349023282635334,
      "grad_norm": 0.20877442927082362,
      "learning_rate": 6.214186108188364e-05,
      "loss": 1.0496,
      "step": 8808
    },
    {
      "epoch": 0.6349744107258704,
      "grad_norm": 0.19184413701921166,
      "learning_rate": 6.212025273780815e-05,
      "loss": 1.0366,
      "step": 8809
    },
    {
      "epoch": 0.6350464931882073,
      "grad_norm": 0.16018887401269502,
      "learning_rate": 6.209864645851777e-05,
      "loss": 1.1071,
      "step": 8810
    },
    {
      "epoch": 0.6351185756505442,
      "grad_norm": 0.1446715610288403,
      "learning_rate": 6.207704224519022e-05,
      "loss": 1.0483,
      "step": 8811
    },
    {
      "epoch": 0.6351906581128811,
      "grad_norm": 0.11832390474809922,
      "learning_rate": 6.205544009900311e-05,
      "loss": 1.0201,
      "step": 8812
    },
    {
      "epoch": 0.635262740575218,
      "grad_norm": 0.11964533074402271,
      "learning_rate": 6.2033840021134e-05,
      "loss": 0.9905,
      "step": 8813
    },
    {
      "epoch": 0.635334823037555,
      "grad_norm": 0.12398095716156049,
      "learning_rate": 6.201224201276028e-05,
      "loss": 1.1146,
      "step": 8814
    },
    {
      "epoch": 0.6354069054998919,
      "grad_norm": 0.1152478948163813,
      "learning_rate": 6.199064607505922e-05,
      "loss": 1.1091,
      "step": 8815
    },
    {
      "epoch": 0.6354789879622288,
      "grad_norm": 0.128474608036305,
      "learning_rate": 6.196905220920795e-05,
      "loss": 1.0516,
      "step": 8816
    },
    {
      "epoch": 0.6355510704245657,
      "grad_norm": 0.13180161048012146,
      "learning_rate": 6.19474604163836e-05,
      "loss": 0.9953,
      "step": 8817
    },
    {
      "epoch": 0.6356231528869026,
      "grad_norm": 0.1271120164513874,
      "learning_rate": 6.19258706977631e-05,
      "loss": 1.0763,
      "step": 8818
    },
    {
      "epoch": 0.6356952353492396,
      "grad_norm": 0.11458538319144417,
      "learning_rate": 6.190428305452324e-05,
      "loss": 1.0865,
      "step": 8819
    },
    {
      "epoch": 0.6357673178115765,
      "grad_norm": 0.2729839592950452,
      "learning_rate": 6.188269748784077e-05,
      "loss": 1.0797,
      "step": 8820
    },
    {
      "epoch": 0.6358394002739134,
      "grad_norm": 0.1259483269049244,
      "learning_rate": 6.186111399889229e-05,
      "loss": 1.0565,
      "step": 8821
    },
    {
      "epoch": 0.6359114827362503,
      "grad_norm": 0.1990309617728338,
      "learning_rate": 6.183953258885433e-05,
      "loss": 1.0237,
      "step": 8822
    },
    {
      "epoch": 0.6359835651985872,
      "grad_norm": 0.11928667566383772,
      "learning_rate": 6.18179532589032e-05,
      "loss": 1.0939,
      "step": 8823
    },
    {
      "epoch": 0.6360556476609242,
      "grad_norm": 0.11043399902284258,
      "learning_rate": 6.179637601021522e-05,
      "loss": 1.0272,
      "step": 8824
    },
    {
      "epoch": 0.636127730123261,
      "grad_norm": 0.173699753284881,
      "learning_rate": 6.17748008439665e-05,
      "loss": 1.0438,
      "step": 8825
    },
    {
      "epoch": 0.6361998125855979,
      "grad_norm": 0.27183145859541974,
      "learning_rate": 6.175322776133314e-05,
      "loss": 1.0322,
      "step": 8826
    },
    {
      "epoch": 0.6362718950479348,
      "grad_norm": 0.12458650772709354,
      "learning_rate": 6.173165676349103e-05,
      "loss": 1.0382,
      "step": 8827
    },
    {
      "epoch": 0.6363439775102717,
      "grad_norm": 0.13537693794710046,
      "learning_rate": 6.171008785161599e-05,
      "loss": 1.1261,
      "step": 8828
    },
    {
      "epoch": 0.6364160599726086,
      "grad_norm": 0.1144973751508234,
      "learning_rate": 6.168852102688372e-05,
      "loss": 1.0683,
      "step": 8829
    },
    {
      "epoch": 0.6364881424349456,
      "grad_norm": 0.6165764795487326,
      "learning_rate": 6.166695629046975e-05,
      "loss": 0.9665,
      "step": 8830
    },
    {
      "epoch": 0.6365602248972825,
      "grad_norm": 0.10965971213037791,
      "learning_rate": 6.164539364354962e-05,
      "loss": 1.112,
      "step": 8831
    },
    {
      "epoch": 0.6366323073596194,
      "grad_norm": 0.11382849451526404,
      "learning_rate": 6.162383308729869e-05,
      "loss": 1.0698,
      "step": 8832
    },
    {
      "epoch": 0.6367043898219563,
      "grad_norm": 0.15757516510076539,
      "learning_rate": 6.160227462289217e-05,
      "loss": 1.0501,
      "step": 8833
    },
    {
      "epoch": 0.6367764722842932,
      "grad_norm": 0.11781402446853431,
      "learning_rate": 6.158071825150518e-05,
      "loss": 0.9918,
      "step": 8834
    },
    {
      "epoch": 0.6368485547466302,
      "grad_norm": 0.1237315943328572,
      "learning_rate": 6.155916397431276e-05,
      "loss": 1.0432,
      "step": 8835
    },
    {
      "epoch": 0.6369206372089671,
      "grad_norm": 0.1450306549555972,
      "learning_rate": 6.153761179248983e-05,
      "loss": 1.0988,
      "step": 8836
    },
    {
      "epoch": 0.636992719671304,
      "grad_norm": 0.12562123314694695,
      "learning_rate": 6.151606170721113e-05,
      "loss": 0.9884,
      "step": 8837
    },
    {
      "epoch": 0.6370648021336409,
      "grad_norm": 0.11914779241516474,
      "learning_rate": 6.149451371965136e-05,
      "loss": 0.9881,
      "step": 8838
    },
    {
      "epoch": 0.6371368845959778,
      "grad_norm": 0.11318990285348543,
      "learning_rate": 6.147296783098506e-05,
      "loss": 1.0206,
      "step": 8839
    },
    {
      "epoch": 0.6372089670583148,
      "grad_norm": 0.1287934743935589,
      "learning_rate": 6.14514240423867e-05,
      "loss": 1.0206,
      "step": 8840
    },
    {
      "epoch": 0.6372810495206517,
      "grad_norm": 0.1036569681482152,
      "learning_rate": 6.14298823550306e-05,
      "loss": 0.9917,
      "step": 8841
    },
    {
      "epoch": 0.6373531319829885,
      "grad_norm": 0.12031709501695717,
      "learning_rate": 6.140834277009098e-05,
      "loss": 1.0277,
      "step": 8842
    },
    {
      "epoch": 0.6374252144453254,
      "grad_norm": 0.12650127424570043,
      "learning_rate": 6.138680528874188e-05,
      "loss": 1.0485,
      "step": 8843
    },
    {
      "epoch": 0.6374972969076623,
      "grad_norm": 0.1309118762866324,
      "learning_rate": 6.136526991215737e-05,
      "loss": 1.0225,
      "step": 8844
    },
    {
      "epoch": 0.6375693793699992,
      "grad_norm": 0.13207876666372145,
      "learning_rate": 6.13437366415113e-05,
      "loss": 1.1049,
      "step": 8845
    },
    {
      "epoch": 0.6376414618323362,
      "grad_norm": 0.18895726741821933,
      "learning_rate": 6.132220547797739e-05,
      "loss": 1.0124,
      "step": 8846
    },
    {
      "epoch": 0.6377135442946731,
      "grad_norm": 0.1257922909454019,
      "learning_rate": 6.130067642272928e-05,
      "loss": 1.154,
      "step": 8847
    },
    {
      "epoch": 0.63778562675701,
      "grad_norm": 0.11552966410517941,
      "learning_rate": 6.127914947694056e-05,
      "loss": 0.9935,
      "step": 8848
    },
    {
      "epoch": 0.6378577092193469,
      "grad_norm": 0.12376936312029117,
      "learning_rate": 6.12576246417846e-05,
      "loss": 1.0214,
      "step": 8849
    },
    {
      "epoch": 0.6379297916816838,
      "grad_norm": 0.11970132959805814,
      "learning_rate": 6.123610191843468e-05,
      "loss": 1.0515,
      "step": 8850
    },
    {
      "epoch": 0.6380018741440208,
      "grad_norm": 0.12245460884280658,
      "learning_rate": 6.121458130806402e-05,
      "loss": 1.0356,
      "step": 8851
    },
    {
      "epoch": 0.6380739566063577,
      "grad_norm": 0.14011933205634447,
      "learning_rate": 6.11930628118456e-05,
      "loss": 0.9625,
      "step": 8852
    },
    {
      "epoch": 0.6381460390686946,
      "grad_norm": 0.29492190806773755,
      "learning_rate": 6.11715464309525e-05,
      "loss": 1.1285,
      "step": 8853
    },
    {
      "epoch": 0.6382181215310315,
      "grad_norm": 0.15729703267523032,
      "learning_rate": 6.115003216655746e-05,
      "loss": 0.9595,
      "step": 8854
    },
    {
      "epoch": 0.6382902039933684,
      "grad_norm": 0.4205712803805889,
      "learning_rate": 6.112852001983325e-05,
      "loss": 1.0724,
      "step": 8855
    },
    {
      "epoch": 0.6383622864557054,
      "grad_norm": 0.1372825917836114,
      "learning_rate": 6.11070099919524e-05,
      "loss": 1.0414,
      "step": 8856
    },
    {
      "epoch": 0.6384343689180423,
      "grad_norm": 0.1398998340110524,
      "learning_rate": 6.108550208408752e-05,
      "loss": 1.122,
      "step": 8857
    },
    {
      "epoch": 0.6385064513803792,
      "grad_norm": 0.13816091218747698,
      "learning_rate": 6.106399629741087e-05,
      "loss": 0.9914,
      "step": 8858
    },
    {
      "epoch": 0.638578533842716,
      "grad_norm": 0.12761385302259223,
      "learning_rate": 6.104249263309479e-05,
      "loss": 1.0525,
      "step": 8859
    },
    {
      "epoch": 0.6386506163050529,
      "grad_norm": 0.2836410839307084,
      "learning_rate": 6.1020991092311364e-05,
      "loss": 1.0292,
      "step": 8860
    },
    {
      "epoch": 0.6387226987673899,
      "grad_norm": 0.11866666374898394,
      "learning_rate": 6.099949167623262e-05,
      "loss": 0.9934,
      "step": 8861
    },
    {
      "epoch": 0.6387947812297268,
      "grad_norm": 0.1458930437297134,
      "learning_rate": 6.097799438603051e-05,
      "loss": 1.079,
      "step": 8862
    },
    {
      "epoch": 0.6388668636920637,
      "grad_norm": 0.5323551112487489,
      "learning_rate": 6.095649922287683e-05,
      "loss": 1.0291,
      "step": 8863
    },
    {
      "epoch": 0.6389389461544006,
      "grad_norm": 0.1289385420953771,
      "learning_rate": 6.09350061879432e-05,
      "loss": 1.0359,
      "step": 8864
    },
    {
      "epoch": 0.6390110286167375,
      "grad_norm": 0.1211493675887127,
      "learning_rate": 6.091351528240121e-05,
      "loss": 1.0484,
      "step": 8865
    },
    {
      "epoch": 0.6390831110790745,
      "grad_norm": 0.12098323410165863,
      "learning_rate": 6.089202650742233e-05,
      "loss": 1.0978,
      "step": 8866
    },
    {
      "epoch": 0.6391551935414114,
      "grad_norm": 0.11755513874183965,
      "learning_rate": 6.08705398641779e-05,
      "loss": 1.015,
      "step": 8867
    },
    {
      "epoch": 0.6392272760037483,
      "grad_norm": 0.12460781348803703,
      "learning_rate": 6.084905535383908e-05,
      "loss": 1.0027,
      "step": 8868
    },
    {
      "epoch": 0.6392993584660852,
      "grad_norm": 0.21667166813249358,
      "learning_rate": 6.0827572977576984e-05,
      "loss": 1.0457,
      "step": 8869
    },
    {
      "epoch": 0.6393714409284221,
      "grad_norm": 0.1117396140729159,
      "learning_rate": 6.0806092736562624e-05,
      "loss": 0.9828,
      "step": 8870
    },
    {
      "epoch": 0.639443523390759,
      "grad_norm": 0.11694703664202677,
      "learning_rate": 6.0784614631966876e-05,
      "loss": 1.0887,
      "step": 8871
    },
    {
      "epoch": 0.639515605853096,
      "grad_norm": 0.13853675119976602,
      "learning_rate": 6.076313866496044e-05,
      "loss": 1.0609,
      "step": 8872
    },
    {
      "epoch": 0.6395876883154329,
      "grad_norm": 0.1642637184374739,
      "learning_rate": 6.074166483671399e-05,
      "loss": 1.0114,
      "step": 8873
    },
    {
      "epoch": 0.6396597707777698,
      "grad_norm": 0.13500370925516636,
      "learning_rate": 6.072019314839795e-05,
      "loss": 1.0073,
      "step": 8874
    },
    {
      "epoch": 0.6397318532401067,
      "grad_norm": 0.11476992693067589,
      "learning_rate": 6.069872360118286e-05,
      "loss": 1.0201,
      "step": 8875
    },
    {
      "epoch": 0.6398039357024436,
      "grad_norm": 0.11780927608295684,
      "learning_rate": 6.067725619623892e-05,
      "loss": 1.0616,
      "step": 8876
    },
    {
      "epoch": 0.6398760181647805,
      "grad_norm": 0.1214285836861607,
      "learning_rate": 6.065579093473631e-05,
      "loss": 0.9922,
      "step": 8877
    },
    {
      "epoch": 0.6399481006271174,
      "grad_norm": 0.12137824604820163,
      "learning_rate": 6.0634327817845036e-05,
      "loss": 0.9974,
      "step": 8878
    },
    {
      "epoch": 0.6400201830894543,
      "grad_norm": 0.11374747769880275,
      "learning_rate": 6.0612866846735114e-05,
      "loss": 1.0029,
      "step": 8879
    },
    {
      "epoch": 0.6400922655517912,
      "grad_norm": 0.1468125427339003,
      "learning_rate": 6.0591408022576314e-05,
      "loss": 1.066,
      "step": 8880
    },
    {
      "epoch": 0.6401643480141281,
      "grad_norm": 0.1544698613863878,
      "learning_rate": 6.056995134653835e-05,
      "loss": 0.9716,
      "step": 8881
    },
    {
      "epoch": 0.6402364304764651,
      "grad_norm": 0.13889120709383695,
      "learning_rate": 6.0548496819790755e-05,
      "loss": 1.037,
      "step": 8882
    },
    {
      "epoch": 0.640308512938802,
      "grad_norm": 1.3325378014897578,
      "learning_rate": 6.052704444350302e-05,
      "loss": 1.0537,
      "step": 8883
    },
    {
      "epoch": 0.6403805954011389,
      "grad_norm": 0.12282132813056806,
      "learning_rate": 6.050559421884452e-05,
      "loss": 0.9396,
      "step": 8884
    },
    {
      "epoch": 0.6404526778634758,
      "grad_norm": 0.3880181740265431,
      "learning_rate": 6.048414614698448e-05,
      "loss": 1.0766,
      "step": 8885
    },
    {
      "epoch": 0.6405247603258127,
      "grad_norm": 0.21616967565611242,
      "learning_rate": 6.046270022909197e-05,
      "loss": 1.0017,
      "step": 8886
    },
    {
      "epoch": 0.6405968427881497,
      "grad_norm": 0.16240495072054145,
      "learning_rate": 6.0441256466335995e-05,
      "loss": 1.0196,
      "step": 8887
    },
    {
      "epoch": 0.6406689252504866,
      "grad_norm": 0.1427461347801306,
      "learning_rate": 6.041981485988546e-05,
      "loss": 1.0622,
      "step": 8888
    },
    {
      "epoch": 0.6407410077128235,
      "grad_norm": 0.1338931571431954,
      "learning_rate": 6.039837541090913e-05,
      "loss": 1.043,
      "step": 8889
    },
    {
      "epoch": 0.6408130901751604,
      "grad_norm": 0.1288276117257511,
      "learning_rate": 6.037693812057561e-05,
      "loss": 1.1036,
      "step": 8890
    },
    {
      "epoch": 0.6408851726374973,
      "grad_norm": 0.12174688086649559,
      "learning_rate": 6.035550299005342e-05,
      "loss": 1.0434,
      "step": 8891
    },
    {
      "epoch": 0.6409572550998343,
      "grad_norm": 0.10923237647312597,
      "learning_rate": 6.033407002051099e-05,
      "loss": 1.0796,
      "step": 8892
    },
    {
      "epoch": 0.6410293375621712,
      "grad_norm": 0.16135923714319883,
      "learning_rate": 6.031263921311665e-05,
      "loss": 1.0143,
      "step": 8893
    },
    {
      "epoch": 0.641101420024508,
      "grad_norm": 0.11929414213206412,
      "learning_rate": 6.029121056903849e-05,
      "loss": 1.035,
      "step": 8894
    },
    {
      "epoch": 0.6411735024868449,
      "grad_norm": 0.13845656899431186,
      "learning_rate": 6.0269784089444605e-05,
      "loss": 1.0052,
      "step": 8895
    },
    {
      "epoch": 0.6412455849491818,
      "grad_norm": 0.1769379376846605,
      "learning_rate": 6.0248359775502904e-05,
      "loss": 0.9855,
      "step": 8896
    },
    {
      "epoch": 0.6413176674115187,
      "grad_norm": 0.11021786812211895,
      "learning_rate": 6.0226937628381264e-05,
      "loss": 0.9832,
      "step": 8897
    },
    {
      "epoch": 0.6413897498738557,
      "grad_norm": 0.1583926976456939,
      "learning_rate": 6.020551764924732e-05,
      "loss": 1.0235,
      "step": 8898
    },
    {
      "epoch": 0.6414618323361926,
      "grad_norm": 0.12236722032197146,
      "learning_rate": 6.018409983926867e-05,
      "loss": 1.0679,
      "step": 8899
    },
    {
      "epoch": 0.6415339147985295,
      "grad_norm": 0.17187227288991372,
      "learning_rate": 6.0162684199612776e-05,
      "loss": 1.0213,
      "step": 8900
    },
    {
      "epoch": 0.6416059972608664,
      "grad_norm": 0.16294724855896808,
      "learning_rate": 6.014127073144702e-05,
      "loss": 1.0884,
      "step": 8901
    },
    {
      "epoch": 0.6416780797232033,
      "grad_norm": 0.12712964271302768,
      "learning_rate": 6.0119859435938586e-05,
      "loss": 1.0156,
      "step": 8902
    },
    {
      "epoch": 0.6417501621855403,
      "grad_norm": 0.13007326426506763,
      "learning_rate": 6.009845031425459e-05,
      "loss": 1.0718,
      "step": 8903
    },
    {
      "epoch": 0.6418222446478772,
      "grad_norm": 0.13907187292066497,
      "learning_rate": 6.007704336756201e-05,
      "loss": 1.0551,
      "step": 8904
    },
    {
      "epoch": 0.6418943271102141,
      "grad_norm": 0.12764622784742852,
      "learning_rate": 6.005563859702772e-05,
      "loss": 1.0388,
      "step": 8905
    },
    {
      "epoch": 0.641966409572551,
      "grad_norm": 0.11742405889259361,
      "learning_rate": 6.0034236003818486e-05,
      "loss": 1.111,
      "step": 8906
    },
    {
      "epoch": 0.6420384920348879,
      "grad_norm": 0.12925273684060304,
      "learning_rate": 6.001283558910094e-05,
      "loss": 1.009,
      "step": 8907
    },
    {
      "epoch": 0.6421105744972249,
      "grad_norm": 0.12965811675590488,
      "learning_rate": 5.9991437354041616e-05,
      "loss": 1.0564,
      "step": 8908
    },
    {
      "epoch": 0.6421826569595618,
      "grad_norm": 0.12528183828589373,
      "learning_rate": 5.997004129980683e-05,
      "loss": 1.1109,
      "step": 8909
    },
    {
      "epoch": 0.6422547394218987,
      "grad_norm": 0.1226656817679445,
      "learning_rate": 5.994864742756293e-05,
      "loss": 1.0519,
      "step": 8910
    },
    {
      "epoch": 0.6423268218842355,
      "grad_norm": 0.1123614115369592,
      "learning_rate": 5.992725573847606e-05,
      "loss": 0.9733,
      "step": 8911
    },
    {
      "epoch": 0.6423989043465724,
      "grad_norm": 0.3777028644209401,
      "learning_rate": 5.990586623371228e-05,
      "loss": 1.0795,
      "step": 8912
    },
    {
      "epoch": 0.6424709868089094,
      "grad_norm": 0.1190965811581948,
      "learning_rate": 5.9884478914437436e-05,
      "loss": 1.0874,
      "step": 8913
    },
    {
      "epoch": 0.6425430692712463,
      "grad_norm": 0.1498892078308829,
      "learning_rate": 5.9863093781817394e-05,
      "loss": 0.955,
      "step": 8914
    },
    {
      "epoch": 0.6426151517335832,
      "grad_norm": 0.12836515457376968,
      "learning_rate": 5.984171083701782e-05,
      "loss": 1.0742,
      "step": 8915
    },
    {
      "epoch": 0.6426872341959201,
      "grad_norm": 0.11293596488011907,
      "learning_rate": 5.982033008120429e-05,
      "loss": 1.051,
      "step": 8916
    },
    {
      "epoch": 0.642759316658257,
      "grad_norm": 0.12828345753091241,
      "learning_rate": 5.97989515155422e-05,
      "loss": 1.013,
      "step": 8917
    },
    {
      "epoch": 0.642831399120594,
      "grad_norm": 0.13792791629665557,
      "learning_rate": 5.977757514119688e-05,
      "loss": 1.0284,
      "step": 8918
    },
    {
      "epoch": 0.6429034815829309,
      "grad_norm": 0.13014362662099635,
      "learning_rate": 5.9756200959333594e-05,
      "loss": 1.0932,
      "step": 8919
    },
    {
      "epoch": 0.6429755640452678,
      "grad_norm": 0.1446853591522466,
      "learning_rate": 5.97348289711174e-05,
      "loss": 1.0723,
      "step": 8920
    },
    {
      "epoch": 0.6430476465076047,
      "grad_norm": 0.11684597684738356,
      "learning_rate": 5.971345917771323e-05,
      "loss": 0.9863,
      "step": 8921
    },
    {
      "epoch": 0.6431197289699416,
      "grad_norm": 0.13662164339635713,
      "learning_rate": 5.969209158028593e-05,
      "loss": 1.0652,
      "step": 8922
    },
    {
      "epoch": 0.6431918114322785,
      "grad_norm": 0.1241544211973251,
      "learning_rate": 5.967072618000025e-05,
      "loss": 1.053,
      "step": 8923
    },
    {
      "epoch": 0.6432638938946155,
      "grad_norm": 0.12392423905747717,
      "learning_rate": 5.964936297802082e-05,
      "loss": 1.0252,
      "step": 8924
    },
    {
      "epoch": 0.6433359763569524,
      "grad_norm": 0.18632415851539422,
      "learning_rate": 5.962800197551206e-05,
      "loss": 1.1458,
      "step": 8925
    },
    {
      "epoch": 0.6434080588192893,
      "grad_norm": 0.17187466500332468,
      "learning_rate": 5.960664317363841e-05,
      "loss": 1.0849,
      "step": 8926
    },
    {
      "epoch": 0.6434801412816262,
      "grad_norm": 0.11190681103114086,
      "learning_rate": 5.958528657356401e-05,
      "loss": 0.9971,
      "step": 8927
    },
    {
      "epoch": 0.6435522237439631,
      "grad_norm": 0.23065291368074134,
      "learning_rate": 5.9563932176453106e-05,
      "loss": 1.0397,
      "step": 8928
    },
    {
      "epoch": 0.6436243062063,
      "grad_norm": 0.12030620230624578,
      "learning_rate": 5.954257998346964e-05,
      "loss": 1.026,
      "step": 8929
    },
    {
      "epoch": 0.6436963886686369,
      "grad_norm": 0.13633773128699772,
      "learning_rate": 5.952122999577752e-05,
      "loss": 1.0537,
      "step": 8930
    },
    {
      "epoch": 0.6437684711309738,
      "grad_norm": 0.4479022643519791,
      "learning_rate": 5.9499882214540435e-05,
      "loss": 0.9902,
      "step": 8931
    },
    {
      "epoch": 0.6438405535933107,
      "grad_norm": 0.1361979074371322,
      "learning_rate": 5.9478536640922165e-05,
      "loss": 0.9004,
      "step": 8932
    },
    {
      "epoch": 0.6439126360556476,
      "grad_norm": 0.12031025796112033,
      "learning_rate": 5.9457193276086145e-05,
      "loss": 1.0223,
      "step": 8933
    },
    {
      "epoch": 0.6439847185179846,
      "grad_norm": 0.11182149358455445,
      "learning_rate": 5.94358521211958e-05,
      "loss": 1.0577,
      "step": 8934
    },
    {
      "epoch": 0.6440568009803215,
      "grad_norm": 0.3922172624571995,
      "learning_rate": 5.9414513177414365e-05,
      "loss": 1.0252,
      "step": 8935
    },
    {
      "epoch": 0.6441288834426584,
      "grad_norm": 0.2166322521529524,
      "learning_rate": 5.9393176445905106e-05,
      "loss": 1.0686,
      "step": 8936
    },
    {
      "epoch": 0.6442009659049953,
      "grad_norm": 0.11455018514347347,
      "learning_rate": 5.9371841927830985e-05,
      "loss": 1.0821,
      "step": 8937
    },
    {
      "epoch": 0.6442730483673322,
      "grad_norm": 0.11840026114680767,
      "learning_rate": 5.935050962435498e-05,
      "loss": 1.0096,
      "step": 8938
    },
    {
      "epoch": 0.6443451308296692,
      "grad_norm": 0.117961056445358,
      "learning_rate": 5.9329179536639835e-05,
      "loss": 1.0169,
      "step": 8939
    },
    {
      "epoch": 0.6444172132920061,
      "grad_norm": 0.12680800753580063,
      "learning_rate": 5.9307851665848244e-05,
      "loss": 0.9986,
      "step": 8940
    },
    {
      "epoch": 0.644489295754343,
      "grad_norm": 0.12349676261031112,
      "learning_rate": 5.9286526013142794e-05,
      "loss": 1.0027,
      "step": 8941
    },
    {
      "epoch": 0.6445613782166799,
      "grad_norm": 0.11468621832464802,
      "learning_rate": 5.926520257968593e-05,
      "loss": 1.0669,
      "step": 8942
    },
    {
      "epoch": 0.6446334606790168,
      "grad_norm": 0.14627004621847398,
      "learning_rate": 5.924388136663992e-05,
      "loss": 0.962,
      "step": 8943
    },
    {
      "epoch": 0.6447055431413538,
      "grad_norm": 0.14928580142304018,
      "learning_rate": 5.922256237516699e-05,
      "loss": 1.0115,
      "step": 8944
    },
    {
      "epoch": 0.6447776256036907,
      "grad_norm": 0.26379858216351076,
      "learning_rate": 5.9201245606429215e-05,
      "loss": 1.0286,
      "step": 8945
    },
    {
      "epoch": 0.6448497080660275,
      "grad_norm": 0.18629378751128484,
      "learning_rate": 5.917993106158859e-05,
      "loss": 1.0524,
      "step": 8946
    },
    {
      "epoch": 0.6449217905283644,
      "grad_norm": 0.11874550447318971,
      "learning_rate": 5.915861874180687e-05,
      "loss": 1.0442,
      "step": 8947
    },
    {
      "epoch": 0.6449938729907013,
      "grad_norm": 0.1099309366715088,
      "learning_rate": 5.9137308648245837e-05,
      "loss": 1.0987,
      "step": 8948
    },
    {
      "epoch": 0.6450659554530382,
      "grad_norm": 0.10437005525976367,
      "learning_rate": 5.9116000782066985e-05,
      "loss": 1.0278,
      "step": 8949
    },
    {
      "epoch": 0.6451380379153752,
      "grad_norm": 0.1511065475246384,
      "learning_rate": 5.909469514443191e-05,
      "loss": 1.0163,
      "step": 8950
    },
    {
      "epoch": 0.6452101203777121,
      "grad_norm": 0.1558099325708563,
      "learning_rate": 5.907339173650188e-05,
      "loss": 1.022,
      "step": 8951
    },
    {
      "epoch": 0.645282202840049,
      "grad_norm": 0.21587742453400324,
      "learning_rate": 5.9052090559438166e-05,
      "loss": 1.0062,
      "step": 8952
    },
    {
      "epoch": 0.6453542853023859,
      "grad_norm": 0.1525347567031758,
      "learning_rate": 5.903079161440179e-05,
      "loss": 1.0423,
      "step": 8953
    },
    {
      "epoch": 0.6454263677647228,
      "grad_norm": 0.10916388126827331,
      "learning_rate": 5.9009494902553844e-05,
      "loss": 1.001,
      "step": 8954
    },
    {
      "epoch": 0.6454984502270598,
      "grad_norm": 0.16763194791045805,
      "learning_rate": 5.8988200425055104e-05,
      "loss": 1.0841,
      "step": 8955
    },
    {
      "epoch": 0.6455705326893967,
      "grad_norm": 0.3388431055574137,
      "learning_rate": 5.896690818306638e-05,
      "loss": 1.0456,
      "step": 8956
    },
    {
      "epoch": 0.6456426151517336,
      "grad_norm": 0.11554250853869288,
      "learning_rate": 5.894561817774819e-05,
      "loss": 1.0766,
      "step": 8957
    },
    {
      "epoch": 0.6457146976140705,
      "grad_norm": 0.13008130872248622,
      "learning_rate": 5.892433041026115e-05,
      "loss": 1.0614,
      "step": 8958
    },
    {
      "epoch": 0.6457867800764074,
      "grad_norm": 0.12721640949855123,
      "learning_rate": 5.890304488176557e-05,
      "loss": 1.0597,
      "step": 8959
    },
    {
      "epoch": 0.6458588625387444,
      "grad_norm": 0.12328025779331628,
      "learning_rate": 5.888176159342171e-05,
      "loss": 0.9872,
      "step": 8960
    },
    {
      "epoch": 0.6459309450010813,
      "grad_norm": 0.3315985191848048,
      "learning_rate": 5.8860480546389684e-05,
      "loss": 1.0431,
      "step": 8961
    },
    {
      "epoch": 0.6460030274634182,
      "grad_norm": 0.13048323114198307,
      "learning_rate": 5.8839201741829483e-05,
      "loss": 1.0217,
      "step": 8962
    },
    {
      "epoch": 0.646075109925755,
      "grad_norm": 0.2480237900516433,
      "learning_rate": 5.881792518090106e-05,
      "loss": 0.9855,
      "step": 8963
    },
    {
      "epoch": 0.6461471923880919,
      "grad_norm": 0.11925465371149373,
      "learning_rate": 5.8796650864764154e-05,
      "loss": 1.0363,
      "step": 8964
    },
    {
      "epoch": 0.6462192748504288,
      "grad_norm": 0.11354356555747067,
      "learning_rate": 5.877537879457837e-05,
      "loss": 1.065,
      "step": 8965
    },
    {
      "epoch": 0.6462913573127658,
      "grad_norm": 0.14145530624123542,
      "learning_rate": 5.875410897150322e-05,
      "loss": 1.0166,
      "step": 8966
    },
    {
      "epoch": 0.6463634397751027,
      "grad_norm": 0.12976350317508117,
      "learning_rate": 5.873284139669816e-05,
      "loss": 1.043,
      "step": 8967
    },
    {
      "epoch": 0.6464355222374396,
      "grad_norm": 0.1404835859276413,
      "learning_rate": 5.871157607132245e-05,
      "loss": 1.0356,
      "step": 8968
    },
    {
      "epoch": 0.6465076046997765,
      "grad_norm": 0.14915005402038944,
      "learning_rate": 5.869031299653518e-05,
      "loss": 1.0592,
      "step": 8969
    },
    {
      "epoch": 0.6465796871621134,
      "grad_norm": 0.13332293038615745,
      "learning_rate": 5.866905217349544e-05,
      "loss": 1.0563,
      "step": 8970
    },
    {
      "epoch": 0.6466517696244504,
      "grad_norm": 0.14116965659007316,
      "learning_rate": 5.864779360336209e-05,
      "loss": 1.0273,
      "step": 8971
    },
    {
      "epoch": 0.6467238520867873,
      "grad_norm": 0.40876573680878187,
      "learning_rate": 5.862653728729397e-05,
      "loss": 1.0389,
      "step": 8972
    },
    {
      "epoch": 0.6467959345491242,
      "grad_norm": 0.10789167371981934,
      "learning_rate": 5.860528322644969e-05,
      "loss": 1.0105,
      "step": 8973
    },
    {
      "epoch": 0.6468680170114611,
      "grad_norm": 0.11490987534538333,
      "learning_rate": 5.858403142198781e-05,
      "loss": 1.0186,
      "step": 8974
    },
    {
      "epoch": 0.646940099473798,
      "grad_norm": 0.19928718001335372,
      "learning_rate": 5.85627818750667e-05,
      "loss": 1.0437,
      "step": 8975
    },
    {
      "epoch": 0.647012181936135,
      "grad_norm": 0.11960350876787543,
      "learning_rate": 5.854153458684473e-05,
      "loss": 1.1083,
      "step": 8976
    },
    {
      "epoch": 0.6470842643984719,
      "grad_norm": 0.14873611405890014,
      "learning_rate": 5.852028955848001e-05,
      "loss": 1.069,
      "step": 8977
    },
    {
      "epoch": 0.6471563468608088,
      "grad_norm": 0.11853444907758018,
      "learning_rate": 5.8499046791130585e-05,
      "loss": 0.9736,
      "step": 8978
    },
    {
      "epoch": 0.6472284293231457,
      "grad_norm": 0.1478591065662922,
      "learning_rate": 5.847780628595439e-05,
      "loss": 1.0137,
      "step": 8979
    },
    {
      "epoch": 0.6473005117854826,
      "grad_norm": 0.15044478899710526,
      "learning_rate": 5.845656804410924e-05,
      "loss": 1.0748,
      "step": 8980
    },
    {
      "epoch": 0.6473725942478195,
      "grad_norm": 0.1267606191457754,
      "learning_rate": 5.843533206675277e-05,
      "loss": 1.0125,
      "step": 8981
    },
    {
      "epoch": 0.6474446767101564,
      "grad_norm": 0.15708185490308238,
      "learning_rate": 5.8414098355042555e-05,
      "loss": 1.0683,
      "step": 8982
    },
    {
      "epoch": 0.6475167591724933,
      "grad_norm": 0.13471402662661253,
      "learning_rate": 5.8392866910136036e-05,
      "loss": 1.0249,
      "step": 8983
    },
    {
      "epoch": 0.6475888416348302,
      "grad_norm": 0.13456419228248462,
      "learning_rate": 5.837163773319046e-05,
      "loss": 0.9836,
      "step": 8984
    },
    {
      "epoch": 0.6476609240971671,
      "grad_norm": 0.12101756548681786,
      "learning_rate": 5.835041082536308e-05,
      "loss": 1.0614,
      "step": 8985
    },
    {
      "epoch": 0.647733006559504,
      "grad_norm": 0.46732389025191867,
      "learning_rate": 5.832918618781089e-05,
      "loss": 1.0485,
      "step": 8986
    },
    {
      "epoch": 0.647805089021841,
      "grad_norm": 0.11498415357152633,
      "learning_rate": 5.830796382169087e-05,
      "loss": 1.064,
      "step": 8987
    },
    {
      "epoch": 0.6478771714841779,
      "grad_norm": 0.14923632975777235,
      "learning_rate": 5.828674372815981e-05,
      "loss": 0.9845,
      "step": 8988
    },
    {
      "epoch": 0.6479492539465148,
      "grad_norm": 0.24099482399166292,
      "learning_rate": 5.8265525908374384e-05,
      "loss": 1.0549,
      "step": 8989
    },
    {
      "epoch": 0.6480213364088517,
      "grad_norm": 0.1960403328283405,
      "learning_rate": 5.8244310363491174e-05,
      "loss": 1.1253,
      "step": 8990
    },
    {
      "epoch": 0.6480934188711887,
      "grad_norm": 0.11580391340708084,
      "learning_rate": 5.82230970946666e-05,
      "loss": 1.1351,
      "step": 8991
    },
    {
      "epoch": 0.6481655013335256,
      "grad_norm": 0.13058546957274172,
      "learning_rate": 5.820188610305701e-05,
      "loss": 1.0239,
      "step": 8992
    },
    {
      "epoch": 0.6482375837958625,
      "grad_norm": 0.13416958784610983,
      "learning_rate": 5.818067738981851e-05,
      "loss": 0.96,
      "step": 8993
    },
    {
      "epoch": 0.6483096662581994,
      "grad_norm": 0.13823113148877697,
      "learning_rate": 5.815947095610723e-05,
      "loss": 1.0504,
      "step": 8994
    },
    {
      "epoch": 0.6483817487205363,
      "grad_norm": 0.1502407781849717,
      "learning_rate": 5.813826680307912e-05,
      "loss": 1.0033,
      "step": 8995
    },
    {
      "epoch": 0.6484538311828733,
      "grad_norm": 0.13747922551019492,
      "learning_rate": 5.811706493189001e-05,
      "loss": 1.0725,
      "step": 8996
    },
    {
      "epoch": 0.6485259136452102,
      "grad_norm": 0.12327737339324264,
      "learning_rate": 5.809586534369547e-05,
      "loss": 1.003,
      "step": 8997
    },
    {
      "epoch": 0.648597996107547,
      "grad_norm": 0.1171453900084453,
      "learning_rate": 5.8074668039651205e-05,
      "loss": 1.0351,
      "step": 8998
    },
    {
      "epoch": 0.6486700785698839,
      "grad_norm": 2.1378234091164576,
      "learning_rate": 5.8053473020912595e-05,
      "loss": 1.0376,
      "step": 8999
    },
    {
      "epoch": 0.6487421610322208,
      "grad_norm": 0.12143993503821891,
      "learning_rate": 5.803228028863501e-05,
      "loss": 0.9735,
      "step": 9000
    },
    {
      "epoch": 0.6488142434945577,
      "grad_norm": 0.12723772985643717,
      "learning_rate": 5.801108984397354e-05,
      "loss": 1.0186,
      "step": 9001
    },
    {
      "epoch": 0.6488863259568947,
      "grad_norm": 0.1260813256944452,
      "learning_rate": 5.798990168808334e-05,
      "loss": 0.9979,
      "step": 9002
    },
    {
      "epoch": 0.6489584084192316,
      "grad_norm": 0.11356290727462162,
      "learning_rate": 5.796871582211934e-05,
      "loss": 1.0638,
      "step": 9003
    },
    {
      "epoch": 0.6490304908815685,
      "grad_norm": 0.319822729192774,
      "learning_rate": 5.794753224723639e-05,
      "loss": 1.0393,
      "step": 9004
    },
    {
      "epoch": 0.6491025733439054,
      "grad_norm": 0.1286524367818307,
      "learning_rate": 5.792635096458909e-05,
      "loss": 1.0753,
      "step": 9005
    },
    {
      "epoch": 0.6491746558062423,
      "grad_norm": 0.11805249150944973,
      "learning_rate": 5.790517197533204e-05,
      "loss": 1.0058,
      "step": 9006
    },
    {
      "epoch": 0.6492467382685793,
      "grad_norm": 0.19431675620317007,
      "learning_rate": 5.788399528061974e-05,
      "loss": 1.0178,
      "step": 9007
    },
    {
      "epoch": 0.6493188207309162,
      "grad_norm": 0.15006654110495257,
      "learning_rate": 5.786282088160653e-05,
      "loss": 1.0217,
      "step": 9008
    },
    {
      "epoch": 0.6493909031932531,
      "grad_norm": 0.14053308832564418,
      "learning_rate": 5.78416487794465e-05,
      "loss": 1.0703,
      "step": 9009
    },
    {
      "epoch": 0.64946298565559,
      "grad_norm": 0.1744801966000109,
      "learning_rate": 5.782047897529373e-05,
      "loss": 1.0148,
      "step": 9010
    },
    {
      "epoch": 0.6495350681179269,
      "grad_norm": 0.11957567917011713,
      "learning_rate": 5.7799311470302255e-05,
      "loss": 0.9848,
      "step": 9011
    },
    {
      "epoch": 0.6496071505802639,
      "grad_norm": 0.14516071026985253,
      "learning_rate": 5.7778146265625874e-05,
      "loss": 1.071,
      "step": 9012
    },
    {
      "epoch": 0.6496792330426008,
      "grad_norm": 0.11844429984352543,
      "learning_rate": 5.7756983362418215e-05,
      "loss": 1.0948,
      "step": 9013
    },
    {
      "epoch": 0.6497513155049377,
      "grad_norm": 0.1354850182317526,
      "learning_rate": 5.773582276183287e-05,
      "loss": 1.0233,
      "step": 9014
    },
    {
      "epoch": 0.6498233979672745,
      "grad_norm": 0.1347365649336514,
      "learning_rate": 5.771466446502326e-05,
      "loss": 1.0625,
      "step": 9015
    },
    {
      "epoch": 0.6498954804296114,
      "grad_norm": 0.14282328756134105,
      "learning_rate": 5.769350847314281e-05,
      "loss": 1.0548,
      "step": 9016
    },
    {
      "epoch": 0.6499675628919483,
      "grad_norm": 0.12870997536087136,
      "learning_rate": 5.767235478734458e-05,
      "loss": 1.1191,
      "step": 9017
    },
    {
      "epoch": 0.6500396453542853,
      "grad_norm": 0.12326014197389715,
      "learning_rate": 5.7651203408781715e-05,
      "loss": 1.0555,
      "step": 9018
    },
    {
      "epoch": 0.6501117278166222,
      "grad_norm": 0.1172912684724908,
      "learning_rate": 5.763005433860706e-05,
      "loss": 1.0272,
      "step": 9019
    },
    {
      "epoch": 0.6501838102789591,
      "grad_norm": 0.12034251734890601,
      "learning_rate": 5.760890757797359e-05,
      "loss": 1.0006,
      "step": 9020
    },
    {
      "epoch": 0.650255892741296,
      "grad_norm": 0.1337900015902048,
      "learning_rate": 5.7587763128033865e-05,
      "loss": 1.0502,
      "step": 9021
    },
    {
      "epoch": 0.650327975203633,
      "grad_norm": 0.12074203863709741,
      "learning_rate": 5.756662098994048e-05,
      "loss": 0.9772,
      "step": 9022
    },
    {
      "epoch": 0.6504000576659699,
      "grad_norm": 0.12407758492477979,
      "learning_rate": 5.754548116484581e-05,
      "loss": 1.0112,
      "step": 9023
    },
    {
      "epoch": 0.6504721401283068,
      "grad_norm": 0.25909471856055505,
      "learning_rate": 5.7524343653902325e-05,
      "loss": 1.16,
      "step": 9024
    },
    {
      "epoch": 0.6505442225906437,
      "grad_norm": 0.12109242650000103,
      "learning_rate": 5.750320845826206e-05,
      "loss": 1.0673,
      "step": 9025
    },
    {
      "epoch": 0.6506163050529806,
      "grad_norm": 0.11813835874121256,
      "learning_rate": 5.748207557907712e-05,
      "loss": 1.0332,
      "step": 9026
    },
    {
      "epoch": 0.6506883875153175,
      "grad_norm": 0.16393204410175424,
      "learning_rate": 5.7460945017499444e-05,
      "loss": 1.1228,
      "step": 9027
    },
    {
      "epoch": 0.6507604699776545,
      "grad_norm": 0.11484557740985774,
      "learning_rate": 5.743981677468082e-05,
      "loss": 1.0594,
      "step": 9028
    },
    {
      "epoch": 0.6508325524399914,
      "grad_norm": 0.1888095821267945,
      "learning_rate": 5.741869085177295e-05,
      "loss": 1.0987,
      "step": 9029
    },
    {
      "epoch": 0.6509046349023283,
      "grad_norm": 0.1249379190467913,
      "learning_rate": 5.739756724992736e-05,
      "loss": 1.0274,
      "step": 9030
    },
    {
      "epoch": 0.6509767173646652,
      "grad_norm": 0.12086236977434547,
      "learning_rate": 5.73764459702955e-05,
      "loss": 1.1069,
      "step": 9031
    },
    {
      "epoch": 0.6510487998270021,
      "grad_norm": 0.15290410713566582,
      "learning_rate": 5.735532701402865e-05,
      "loss": 1.0564,
      "step": 9032
    },
    {
      "epoch": 0.651120882289339,
      "grad_norm": 0.1875609908055292,
      "learning_rate": 5.7334210382277975e-05,
      "loss": 1.0667,
      "step": 9033
    },
    {
      "epoch": 0.6511929647516759,
      "grad_norm": 0.14379111130292863,
      "learning_rate": 5.731309607619455e-05,
      "loss": 1.0603,
      "step": 9034
    },
    {
      "epoch": 0.6512650472140128,
      "grad_norm": 0.14191377062114074,
      "learning_rate": 5.729198409692926e-05,
      "loss": 1.0529,
      "step": 9035
    },
    {
      "epoch": 0.6513371296763497,
      "grad_norm": 0.1211275775136108,
      "learning_rate": 5.727087444563293e-05,
      "loss": 1.0602,
      "step": 9036
    },
    {
      "epoch": 0.6514092121386866,
      "grad_norm": 0.11697269912753808,
      "learning_rate": 5.72497671234562e-05,
      "loss": 1.0012,
      "step": 9037
    },
    {
      "epoch": 0.6514812946010236,
      "grad_norm": 0.12760770749344752,
      "learning_rate": 5.722866213154962e-05,
      "loss": 1.0422,
      "step": 9038
    },
    {
      "epoch": 0.6515533770633605,
      "grad_norm": 0.13876113933571368,
      "learning_rate": 5.7207559471063596e-05,
      "loss": 1.043,
      "step": 9039
    },
    {
      "epoch": 0.6516254595256974,
      "grad_norm": 0.11496474345653118,
      "learning_rate": 5.718645914314842e-05,
      "loss": 1.0057,
      "step": 9040
    },
    {
      "epoch": 0.6516975419880343,
      "grad_norm": 0.1365138204760935,
      "learning_rate": 5.7165361148954233e-05,
      "loss": 1.0541,
      "step": 9041
    },
    {
      "epoch": 0.6517696244503712,
      "grad_norm": 0.12249912707463924,
      "learning_rate": 5.714426548963108e-05,
      "loss": 1.047,
      "step": 9042
    },
    {
      "epoch": 0.6518417069127082,
      "grad_norm": 0.1236390858517415,
      "learning_rate": 5.712317216632884e-05,
      "loss": 1.0484,
      "step": 9043
    },
    {
      "epoch": 0.6519137893750451,
      "grad_norm": 0.11801022743526765,
      "learning_rate": 5.710208118019732e-05,
      "loss": 0.9913,
      "step": 9044
    },
    {
      "epoch": 0.651985871837382,
      "grad_norm": 0.11489069030987491,
      "learning_rate": 5.708099253238614e-05,
      "loss": 1.0518,
      "step": 9045
    },
    {
      "epoch": 0.6520579542997189,
      "grad_norm": 0.7262830961077568,
      "learning_rate": 5.7059906224044845e-05,
      "loss": 1.0828,
      "step": 9046
    },
    {
      "epoch": 0.6521300367620558,
      "grad_norm": 0.11019888452058726,
      "learning_rate": 5.70388222563228e-05,
      "loss": 1.0394,
      "step": 9047
    },
    {
      "epoch": 0.6522021192243928,
      "grad_norm": 0.14932523229789127,
      "learning_rate": 5.701774063036929e-05,
      "loss": 1.0466,
      "step": 9048
    },
    {
      "epoch": 0.6522742016867297,
      "grad_norm": 0.11854630012988206,
      "learning_rate": 5.699666134733349e-05,
      "loss": 1.0799,
      "step": 9049
    },
    {
      "epoch": 0.6523462841490665,
      "grad_norm": 0.13724675203556727,
      "learning_rate": 5.697558440836428e-05,
      "loss": 1.0806,
      "step": 9050
    },
    {
      "epoch": 0.6524183666114034,
      "grad_norm": 0.12513912482883682,
      "learning_rate": 5.695450981461067e-05,
      "loss": 1.0252,
      "step": 9051
    },
    {
      "epoch": 0.6524904490737403,
      "grad_norm": 0.11829219704019904,
      "learning_rate": 5.6933437567221374e-05,
      "loss": 1.0122,
      "step": 9052
    },
    {
      "epoch": 0.6525625315360772,
      "grad_norm": 0.1261446238963491,
      "learning_rate": 5.691236766734505e-05,
      "loss": 1.021,
      "step": 9053
    },
    {
      "epoch": 0.6526346139984142,
      "grad_norm": 0.1190410962714567,
      "learning_rate": 5.6891300116130076e-05,
      "loss": 1.0905,
      "step": 9054
    },
    {
      "epoch": 0.6527066964607511,
      "grad_norm": 0.21953317418979248,
      "learning_rate": 5.6870234914724964e-05,
      "loss": 1.0242,
      "step": 9055
    },
    {
      "epoch": 0.652778778923088,
      "grad_norm": 0.31869602283324755,
      "learning_rate": 5.6849172064277876e-05,
      "loss": 1.039,
      "step": 9056
    },
    {
      "epoch": 0.6528508613854249,
      "grad_norm": 0.1231151160978406,
      "learning_rate": 5.6828111565937e-05,
      "loss": 1.0153,
      "step": 9057
    },
    {
      "epoch": 0.6529229438477618,
      "grad_norm": 0.11561539003346588,
      "learning_rate": 5.680705342085024e-05,
      "loss": 1.0714,
      "step": 9058
    },
    {
      "epoch": 0.6529950263100988,
      "grad_norm": 0.12758951449688832,
      "learning_rate": 5.6785997630165435e-05,
      "loss": 1.088,
      "step": 9059
    },
    {
      "epoch": 0.6530671087724357,
      "grad_norm": 0.11827998242255165,
      "learning_rate": 5.676494419503044e-05,
      "loss": 0.9602,
      "step": 9060
    },
    {
      "epoch": 0.6531391912347726,
      "grad_norm": 0.1417368080799537,
      "learning_rate": 5.674389311659274e-05,
      "loss": 0.9419,
      "step": 9061
    },
    {
      "epoch": 0.6532112736971095,
      "grad_norm": 0.1144010312144856,
      "learning_rate": 5.672284439599984e-05,
      "loss": 1.0386,
      "step": 9062
    },
    {
      "epoch": 0.6532833561594464,
      "grad_norm": 0.11061065400955283,
      "learning_rate": 5.670179803439906e-05,
      "loss": 0.9866,
      "step": 9063
    },
    {
      "epoch": 0.6533554386217834,
      "grad_norm": 0.10900781491951021,
      "learning_rate": 5.668075403293771e-05,
      "loss": 1.0494,
      "step": 9064
    },
    {
      "epoch": 0.6534275210841203,
      "grad_norm": 0.12522950270454503,
      "learning_rate": 5.665971239276279e-05,
      "loss": 1.0694,
      "step": 9065
    },
    {
      "epoch": 0.6534996035464572,
      "grad_norm": 0.10630572022991702,
      "learning_rate": 5.663867311502127e-05,
      "loss": 1.006,
      "step": 9066
    },
    {
      "epoch": 0.653571686008794,
      "grad_norm": 0.12197270363080157,
      "learning_rate": 5.6617636200859935e-05,
      "loss": 1.0311,
      "step": 9067
    },
    {
      "epoch": 0.6536437684711309,
      "grad_norm": 0.12551368275380068,
      "learning_rate": 5.659660165142563e-05,
      "loss": 1.11,
      "step": 9068
    },
    {
      "epoch": 0.6537158509334678,
      "grad_norm": 0.14062147635538358,
      "learning_rate": 5.6575569467864784e-05,
      "loss": 1.0808,
      "step": 9069
    },
    {
      "epoch": 0.6537879333958048,
      "grad_norm": 0.12733324645777203,
      "learning_rate": 5.655453965132389e-05,
      "loss": 1.0744,
      "step": 9070
    },
    {
      "epoch": 0.6538600158581417,
      "grad_norm": 0.11673549836061824,
      "learning_rate": 5.6533512202949267e-05,
      "loss": 1.0586,
      "step": 9071
    },
    {
      "epoch": 0.6539320983204786,
      "grad_norm": 0.12159067451195694,
      "learning_rate": 5.651248712388708e-05,
      "loss": 1.0164,
      "step": 9072
    },
    {
      "epoch": 0.6540041807828155,
      "grad_norm": 0.23614641295216904,
      "learning_rate": 5.6491464415283416e-05,
      "loss": 1.0099,
      "step": 9073
    },
    {
      "epoch": 0.6540762632451524,
      "grad_norm": 0.12981283155042858,
      "learning_rate": 5.6470444078284155e-05,
      "loss": 0.9934,
      "step": 9074
    },
    {
      "epoch": 0.6541483457074894,
      "grad_norm": 0.12431929511383055,
      "learning_rate": 5.644942611403513e-05,
      "loss": 0.9755,
      "step": 9075
    },
    {
      "epoch": 0.6542204281698263,
      "grad_norm": 0.11244420517905235,
      "learning_rate": 5.642841052368199e-05,
      "loss": 0.9851,
      "step": 9076
    },
    {
      "epoch": 0.6542925106321632,
      "grad_norm": 0.12745798953894102,
      "learning_rate": 5.640739730837028e-05,
      "loss": 1.1287,
      "step": 9077
    },
    {
      "epoch": 0.6543645930945001,
      "grad_norm": 0.10987983648027619,
      "learning_rate": 5.638638646924541e-05,
      "loss": 1.0351,
      "step": 9078
    },
    {
      "epoch": 0.654436675556837,
      "grad_norm": 0.1140537925796729,
      "learning_rate": 5.6365378007452664e-05,
      "loss": 1.0097,
      "step": 9079
    },
    {
      "epoch": 0.654508758019174,
      "grad_norm": 0.12358134449948972,
      "learning_rate": 5.6344371924137173e-05,
      "loss": 0.9446,
      "step": 9080
    },
    {
      "epoch": 0.6545808404815109,
      "grad_norm": 0.16400471903614483,
      "learning_rate": 5.632336822044397e-05,
      "loss": 0.9998,
      "step": 9081
    },
    {
      "epoch": 0.6546529229438478,
      "grad_norm": 0.11623543778845653,
      "learning_rate": 5.6302366897517946e-05,
      "loss": 1.0586,
      "step": 9082
    },
    {
      "epoch": 0.6547250054061847,
      "grad_norm": 0.1298871242499509,
      "learning_rate": 5.628136795650385e-05,
      "loss": 1.091,
      "step": 9083
    },
    {
      "epoch": 0.6547970878685216,
      "grad_norm": 0.11373440598758106,
      "learning_rate": 5.626037139854632e-05,
      "loss": 0.9921,
      "step": 9084
    },
    {
      "epoch": 0.6548691703308585,
      "grad_norm": 0.1176820083263499,
      "learning_rate": 5.6239377224789845e-05,
      "loss": 1.047,
      "step": 9085
    },
    {
      "epoch": 0.6549412527931954,
      "grad_norm": 0.11571604123546327,
      "learning_rate": 5.621838543637883e-05,
      "loss": 1.0605,
      "step": 9086
    },
    {
      "epoch": 0.6550133352555323,
      "grad_norm": 0.14163649177189871,
      "learning_rate": 5.6197396034457464e-05,
      "loss": 1.0198,
      "step": 9087
    },
    {
      "epoch": 0.6550854177178692,
      "grad_norm": 0.12317579687007038,
      "learning_rate": 5.61764090201699e-05,
      "loss": 1.0908,
      "step": 9088
    },
    {
      "epoch": 0.6551575001802061,
      "grad_norm": 0.11227520435144321,
      "learning_rate": 5.615542439466008e-05,
      "loss": 1.0011,
      "step": 9089
    },
    {
      "epoch": 0.655229582642543,
      "grad_norm": 0.14439974828151206,
      "learning_rate": 5.613444215907191e-05,
      "loss": 1.0361,
      "step": 9090
    },
    {
      "epoch": 0.65530166510488,
      "grad_norm": 0.14343987693035684,
      "learning_rate": 5.6113462314549046e-05,
      "loss": 0.9022,
      "step": 9091
    },
    {
      "epoch": 0.6553737475672169,
      "grad_norm": 0.41356007370515535,
      "learning_rate": 5.609248486223512e-05,
      "loss": 1.0983,
      "step": 9092
    },
    {
      "epoch": 0.6554458300295538,
      "grad_norm": 0.11750598182046085,
      "learning_rate": 5.6071509803273604e-05,
      "loss": 1.0345,
      "step": 9093
    },
    {
      "epoch": 0.6555179124918907,
      "grad_norm": 0.11890180341912242,
      "learning_rate": 5.6050537138807723e-05,
      "loss": 1.0707,
      "step": 9094
    },
    {
      "epoch": 0.6555899949542277,
      "grad_norm": 0.11791338618784225,
      "learning_rate": 5.60295668699808e-05,
      "loss": 1.0483,
      "step": 9095
    },
    {
      "epoch": 0.6556620774165646,
      "grad_norm": 0.13887915802078407,
      "learning_rate": 5.600859899793584e-05,
      "loss": 1.0919,
      "step": 9096
    },
    {
      "epoch": 0.6557341598789015,
      "grad_norm": 0.12590370255729894,
      "learning_rate": 5.598763352381585e-05,
      "loss": 1.0542,
      "step": 9097
    },
    {
      "epoch": 0.6558062423412384,
      "grad_norm": 0.16572599897064372,
      "learning_rate": 5.596667044876349e-05,
      "loss": 0.8649,
      "step": 9098
    },
    {
      "epoch": 0.6558783248035753,
      "grad_norm": 0.13636370685958024,
      "learning_rate": 5.5945709773921576e-05,
      "loss": 1.0009,
      "step": 9099
    },
    {
      "epoch": 0.6559504072659122,
      "grad_norm": 0.15963003927322347,
      "learning_rate": 5.5924751500432584e-05,
      "loss": 1.0113,
      "step": 9100
    },
    {
      "epoch": 0.6560224897282492,
      "grad_norm": 0.1340275919213019,
      "learning_rate": 5.590379562943899e-05,
      "loss": 1.0561,
      "step": 9101
    },
    {
      "epoch": 0.656094572190586,
      "grad_norm": 0.12557812953287095,
      "learning_rate": 5.5882842162082995e-05,
      "loss": 0.9961,
      "step": 9102
    },
    {
      "epoch": 0.6561666546529229,
      "grad_norm": 0.2796086633350518,
      "learning_rate": 5.5861891099506723e-05,
      "loss": 1.027,
      "step": 9103
    },
    {
      "epoch": 0.6562387371152598,
      "grad_norm": 0.11942436340320789,
      "learning_rate": 5.5840942442852316e-05,
      "loss": 1.0694,
      "step": 9104
    },
    {
      "epoch": 0.6563108195775967,
      "grad_norm": 0.1372343478297001,
      "learning_rate": 5.581999619326164e-05,
      "loss": 1.0687,
      "step": 9105
    },
    {
      "epoch": 0.6563829020399337,
      "grad_norm": 0.10637047283153558,
      "learning_rate": 5.579905235187639e-05,
      "loss": 1.1077,
      "step": 9106
    },
    {
      "epoch": 0.6564549845022706,
      "grad_norm": 0.14905582773251644,
      "learning_rate": 5.5778110919838154e-05,
      "loss": 1.0533,
      "step": 9107
    },
    {
      "epoch": 0.6565270669646075,
      "grad_norm": 0.1392190950785321,
      "learning_rate": 5.5757171898288554e-05,
      "loss": 1.0807,
      "step": 9108
    },
    {
      "epoch": 0.6565991494269444,
      "grad_norm": 0.13378570979656795,
      "learning_rate": 5.573623528836893e-05,
      "loss": 1.0542,
      "step": 9109
    },
    {
      "epoch": 0.6566712318892813,
      "grad_norm": 0.1331865631962053,
      "learning_rate": 5.571530109122043e-05,
      "loss": 1.0319,
      "step": 9110
    },
    {
      "epoch": 0.6567433143516183,
      "grad_norm": 0.1274381281510231,
      "learning_rate": 5.5694369307984204e-05,
      "loss": 0.9993,
      "step": 9111
    },
    {
      "epoch": 0.6568153968139552,
      "grad_norm": 0.36106984251041263,
      "learning_rate": 5.567343993980119e-05,
      "loss": 1.0876,
      "step": 9112
    },
    {
      "epoch": 0.6568874792762921,
      "grad_norm": 0.14457230171431285,
      "learning_rate": 5.565251298781233e-05,
      "loss": 1.0352,
      "step": 9113
    },
    {
      "epoch": 0.656959561738629,
      "grad_norm": 0.10430568443624742,
      "learning_rate": 5.5631588453158214e-05,
      "loss": 1.0439,
      "step": 9114
    },
    {
      "epoch": 0.6570316442009659,
      "grad_norm": 0.1337900325903131,
      "learning_rate": 5.5610666336979454e-05,
      "loss": 1.0694,
      "step": 9115
    },
    {
      "epoch": 0.6571037266633029,
      "grad_norm": 0.23972649420220996,
      "learning_rate": 5.5589746640416465e-05,
      "loss": 1.0933,
      "step": 9116
    },
    {
      "epoch": 0.6571758091256398,
      "grad_norm": 0.10595377989371114,
      "learning_rate": 5.5568829364609664e-05,
      "loss": 1.0491,
      "step": 9117
    },
    {
      "epoch": 0.6572478915879767,
      "grad_norm": 0.13015184545191522,
      "learning_rate": 5.554791451069912e-05,
      "loss": 1.0715,
      "step": 9118
    },
    {
      "epoch": 0.6573199740503135,
      "grad_norm": 0.12445409503551512,
      "learning_rate": 5.552700207982491e-05,
      "loss": 0.9928,
      "step": 9119
    },
    {
      "epoch": 0.6573920565126504,
      "grad_norm": 0.111980500858824,
      "learning_rate": 5.5506092073126914e-05,
      "loss": 1.0028,
      "step": 9120
    },
    {
      "epoch": 0.6574641389749873,
      "grad_norm": 0.11015101730925511,
      "learning_rate": 5.5485184491745043e-05,
      "loss": 1.0125,
      "step": 9121
    },
    {
      "epoch": 0.6575362214373243,
      "grad_norm": 0.1389246512497652,
      "learning_rate": 5.5464279336818816e-05,
      "loss": 1.0652,
      "step": 9122
    },
    {
      "epoch": 0.6576083038996612,
      "grad_norm": 0.1204909484572866,
      "learning_rate": 5.544337660948779e-05,
      "loss": 1.0662,
      "step": 9123
    },
    {
      "epoch": 0.6576803863619981,
      "grad_norm": 0.1408630744815012,
      "learning_rate": 5.542247631089135e-05,
      "loss": 1.0748,
      "step": 9124
    },
    {
      "epoch": 0.657752468824335,
      "grad_norm": 0.11239116543761003,
      "learning_rate": 5.540157844216876e-05,
      "loss": 1.059,
      "step": 9125
    },
    {
      "epoch": 0.6578245512866719,
      "grad_norm": 0.1529182775353567,
      "learning_rate": 5.5380683004459124e-05,
      "loss": 0.9686,
      "step": 9126
    },
    {
      "epoch": 0.6578966337490089,
      "grad_norm": 0.142531407792427,
      "learning_rate": 5.535978999890144e-05,
      "loss": 1.0134,
      "step": 9127
    },
    {
      "epoch": 0.6579687162113458,
      "grad_norm": 0.13981462825672594,
      "learning_rate": 5.5338899426634575e-05,
      "loss": 1.0309,
      "step": 9128
    },
    {
      "epoch": 0.6580407986736827,
      "grad_norm": 0.11906398725452934,
      "learning_rate": 5.5318011288797237e-05,
      "loss": 1.027,
      "step": 9129
    },
    {
      "epoch": 0.6581128811360196,
      "grad_norm": 0.12128591973500931,
      "learning_rate": 5.529712558652802e-05,
      "loss": 1.0436,
      "step": 9130
    },
    {
      "epoch": 0.6581849635983565,
      "grad_norm": 0.13762332539700806,
      "learning_rate": 5.5276242320965374e-05,
      "loss": 1.0465,
      "step": 9131
    },
    {
      "epoch": 0.6582570460606935,
      "grad_norm": 0.1322395747798928,
      "learning_rate": 5.5255361493247634e-05,
      "loss": 1.1437,
      "step": 9132
    },
    {
      "epoch": 0.6583291285230304,
      "grad_norm": 0.12632934675986288,
      "learning_rate": 5.5234483104513005e-05,
      "loss": 0.9675,
      "step": 9133
    },
    {
      "epoch": 0.6584012109853673,
      "grad_norm": 0.112939821374489,
      "learning_rate": 5.521360715589952e-05,
      "loss": 1.0629,
      "step": 9134
    },
    {
      "epoch": 0.6584732934477042,
      "grad_norm": 0.1750960247854396,
      "learning_rate": 5.5192733648545116e-05,
      "loss": 1.08,
      "step": 9135
    },
    {
      "epoch": 0.6585453759100411,
      "grad_norm": 0.12589374816848778,
      "learning_rate": 5.5171862583587576e-05,
      "loss": 1.0789,
      "step": 9136
    },
    {
      "epoch": 0.658617458372378,
      "grad_norm": 0.14646170518772658,
      "learning_rate": 5.5150993962164574e-05,
      "loss": 0.999,
      "step": 9137
    },
    {
      "epoch": 0.6586895408347149,
      "grad_norm": 0.17288897394600358,
      "learning_rate": 5.513012778541364e-05,
      "loss": 1.0664,
      "step": 9138
    },
    {
      "epoch": 0.6587616232970518,
      "grad_norm": 0.11453478951793021,
      "learning_rate": 5.510926405447215e-05,
      "loss": 1.0454,
      "step": 9139
    },
    {
      "epoch": 0.6588337057593887,
      "grad_norm": 0.12036513587697957,
      "learning_rate": 5.5088402770477376e-05,
      "loss": 1.0065,
      "step": 9140
    },
    {
      "epoch": 0.6589057882217256,
      "grad_norm": 0.12733172789614355,
      "learning_rate": 5.506754393456645e-05,
      "loss": 1.0496,
      "step": 9141
    },
    {
      "epoch": 0.6589778706840625,
      "grad_norm": 0.11467718442668591,
      "learning_rate": 5.504668754787634e-05,
      "loss": 0.9822,
      "step": 9142
    },
    {
      "epoch": 0.6590499531463995,
      "grad_norm": 0.11250723455037433,
      "learning_rate": 5.5025833611543945e-05,
      "loss": 1.0517,
      "step": 9143
    },
    {
      "epoch": 0.6591220356087364,
      "grad_norm": 0.12189740859627458,
      "learning_rate": 5.500498212670595e-05,
      "loss": 1.0889,
      "step": 9144
    },
    {
      "epoch": 0.6591941180710733,
      "grad_norm": 0.11419823416865837,
      "learning_rate": 5.498413309449897e-05,
      "loss": 0.9658,
      "step": 9145
    },
    {
      "epoch": 0.6592662005334102,
      "grad_norm": 0.10912385584669858,
      "learning_rate": 5.4963286516059496e-05,
      "loss": 0.9757,
      "step": 9146
    },
    {
      "epoch": 0.6593382829957471,
      "grad_norm": 0.4326435579418354,
      "learning_rate": 5.4942442392523754e-05,
      "loss": 1.0373,
      "step": 9147
    },
    {
      "epoch": 0.6594103654580841,
      "grad_norm": 0.10936783101393263,
      "learning_rate": 5.492160072502801e-05,
      "loss": 1.0666,
      "step": 9148
    },
    {
      "epoch": 0.659482447920421,
      "grad_norm": 0.10672903306387034,
      "learning_rate": 5.490076151470833e-05,
      "loss": 0.9585,
      "step": 9149
    },
    {
      "epoch": 0.6595545303827579,
      "grad_norm": 0.11388895185604822,
      "learning_rate": 5.4879924762700655e-05,
      "loss": 1.0041,
      "step": 9150
    },
    {
      "epoch": 0.6596266128450948,
      "grad_norm": 0.11607658281558424,
      "learning_rate": 5.485909047014066e-05,
      "loss": 1.0354,
      "step": 9151
    },
    {
      "epoch": 0.6596986953074317,
      "grad_norm": 0.11271103116040888,
      "learning_rate": 5.48382586381641e-05,
      "loss": 0.9808,
      "step": 9152
    },
    {
      "epoch": 0.6597707777697687,
      "grad_norm": 0.12079232293957626,
      "learning_rate": 5.481742926790649e-05,
      "loss": 1.0101,
      "step": 9153
    },
    {
      "epoch": 0.6598428602321055,
      "grad_norm": 0.17091160229780253,
      "learning_rate": 5.4796602360503235e-05,
      "loss": 1.0228,
      "step": 9154
    },
    {
      "epoch": 0.6599149426944424,
      "grad_norm": 0.12348109411984506,
      "learning_rate": 5.47757779170895e-05,
      "loss": 1.0295,
      "step": 9155
    },
    {
      "epoch": 0.6599870251567793,
      "grad_norm": 0.11799835483505608,
      "learning_rate": 5.4754955938800426e-05,
      "loss": 1.076,
      "step": 9156
    },
    {
      "epoch": 0.6600591076191162,
      "grad_norm": 0.12227476294988399,
      "learning_rate": 5.473413642677108e-05,
      "loss": 1.0404,
      "step": 9157
    },
    {
      "epoch": 0.6601311900814532,
      "grad_norm": 0.1174098043679162,
      "learning_rate": 5.4713319382136274e-05,
      "loss": 1.0802,
      "step": 9158
    },
    {
      "epoch": 0.6602032725437901,
      "grad_norm": 0.1802223939491406,
      "learning_rate": 5.4692504806030674e-05,
      "loss": 1.0782,
      "step": 9159
    },
    {
      "epoch": 0.660275355006127,
      "grad_norm": 0.12201729839764719,
      "learning_rate": 5.467169269958887e-05,
      "loss": 1.0491,
      "step": 9160
    },
    {
      "epoch": 0.6603474374684639,
      "grad_norm": 0.13017083153800943,
      "learning_rate": 5.4650883063945366e-05,
      "loss": 1.0617,
      "step": 9161
    },
    {
      "epoch": 0.6604195199308008,
      "grad_norm": 0.13027207400396668,
      "learning_rate": 5.463007590023448e-05,
      "loss": 0.9469,
      "step": 9162
    },
    {
      "epoch": 0.6604916023931378,
      "grad_norm": 0.12509947003043898,
      "learning_rate": 5.460927120959029e-05,
      "loss": 1.086,
      "step": 9163
    },
    {
      "epoch": 0.6605636848554747,
      "grad_norm": 0.12570373141198918,
      "learning_rate": 5.458846899314687e-05,
      "loss": 1.0232,
      "step": 9164
    },
    {
      "epoch": 0.6606357673178116,
      "grad_norm": 0.2690634005119692,
      "learning_rate": 5.45676692520382e-05,
      "loss": 1.0721,
      "step": 9165
    },
    {
      "epoch": 0.6607078497801485,
      "grad_norm": 0.20111192749540074,
      "learning_rate": 5.454687198739804e-05,
      "loss": 1.1587,
      "step": 9166
    },
    {
      "epoch": 0.6607799322424854,
      "grad_norm": 0.14148038141266608,
      "learning_rate": 5.452607720035995e-05,
      "loss": 1.0787,
      "step": 9167
    },
    {
      "epoch": 0.6608520147048224,
      "grad_norm": 0.17044546076025588,
      "learning_rate": 5.4505284892057476e-05,
      "loss": 1.0407,
      "step": 9168
    },
    {
      "epoch": 0.6609240971671593,
      "grad_norm": 0.11914073989312036,
      "learning_rate": 5.448449506362394e-05,
      "loss": 1.0559,
      "step": 9169
    },
    {
      "epoch": 0.6609961796294962,
      "grad_norm": 0.14645152542296774,
      "learning_rate": 5.4463707716192694e-05,
      "loss": 1.1039,
      "step": 9170
    },
    {
      "epoch": 0.661068262091833,
      "grad_norm": 0.12170755915202984,
      "learning_rate": 5.4442922850896726e-05,
      "loss": 1.0758,
      "step": 9171
    },
    {
      "epoch": 0.6611403445541699,
      "grad_norm": 0.11964088269392585,
      "learning_rate": 5.442214046886902e-05,
      "loss": 1.0759,
      "step": 9172
    },
    {
      "epoch": 0.6612124270165068,
      "grad_norm": 0.11850520743841302,
      "learning_rate": 5.4401360571242375e-05,
      "loss": 0.9763,
      "step": 9173
    },
    {
      "epoch": 0.6612845094788438,
      "grad_norm": 0.10942131784430949,
      "learning_rate": 5.438058315914959e-05,
      "loss": 1.0663,
      "step": 9174
    },
    {
      "epoch": 0.6613565919411807,
      "grad_norm": 0.17543496259408534,
      "learning_rate": 5.435980823372311e-05,
      "loss": 0.9655,
      "step": 9175
    },
    {
      "epoch": 0.6614286744035176,
      "grad_norm": 0.11310780143813515,
      "learning_rate": 5.4339035796095386e-05,
      "loss": 0.9788,
      "step": 9176
    },
    {
      "epoch": 0.6615007568658545,
      "grad_norm": 0.10650632866389241,
      "learning_rate": 5.431826584739872e-05,
      "loss": 1.0092,
      "step": 9177
    },
    {
      "epoch": 0.6615728393281914,
      "grad_norm": 0.11103399919594777,
      "learning_rate": 5.4297498388765234e-05,
      "loss": 1.0777,
      "step": 9178
    },
    {
      "epoch": 0.6616449217905284,
      "grad_norm": 0.33184883058878734,
      "learning_rate": 5.427673342132695e-05,
      "loss": 1.0329,
      "step": 9179
    },
    {
      "epoch": 0.6617170042528653,
      "grad_norm": 0.11288355707967869,
      "learning_rate": 5.425597094621575e-05,
      "loss": 1.0113,
      "step": 9180
    },
    {
      "epoch": 0.6617890867152022,
      "grad_norm": 0.13328535718411122,
      "learning_rate": 5.423521096456336e-05,
      "loss": 1.0027,
      "step": 9181
    },
    {
      "epoch": 0.6618611691775391,
      "grad_norm": 0.11443567776653331,
      "learning_rate": 5.421445347750141e-05,
      "loss": 1.0548,
      "step": 9182
    },
    {
      "epoch": 0.661933251639876,
      "grad_norm": 0.12000953913773495,
      "learning_rate": 5.419369848616136e-05,
      "loss": 1.0723,
      "step": 9183
    },
    {
      "epoch": 0.662005334102213,
      "grad_norm": 0.3794732320406507,
      "learning_rate": 5.417294599167453e-05,
      "loss": 1.0421,
      "step": 9184
    },
    {
      "epoch": 0.6620774165645499,
      "grad_norm": 0.12097788764294748,
      "learning_rate": 5.415219599517212e-05,
      "loss": 1.0416,
      "step": 9185
    },
    {
      "epoch": 0.6621494990268868,
      "grad_norm": 0.12641726436159828,
      "learning_rate": 5.4131448497785196e-05,
      "loss": 0.9961,
      "step": 9186
    },
    {
      "epoch": 0.6622215814892237,
      "grad_norm": 0.13609673354264812,
      "learning_rate": 5.4110703500644676e-05,
      "loss": 1.0079,
      "step": 9187
    },
    {
      "epoch": 0.6622936639515606,
      "grad_norm": 0.13592461583784235,
      "learning_rate": 5.408996100488135e-05,
      "loss": 0.9395,
      "step": 9188
    },
    {
      "epoch": 0.6623657464138974,
      "grad_norm": 0.10855368269434489,
      "learning_rate": 5.4069221011625884e-05,
      "loss": 1.0865,
      "step": 9189
    },
    {
      "epoch": 0.6624378288762344,
      "grad_norm": 0.13407451443422536,
      "learning_rate": 5.4048483522008776e-05,
      "loss": 1.129,
      "step": 9190
    },
    {
      "epoch": 0.6625099113385713,
      "grad_norm": 0.15376576184135876,
      "learning_rate": 5.402774853716041e-05,
      "loss": 1.0387,
      "step": 9191
    },
    {
      "epoch": 0.6625819938009082,
      "grad_norm": 0.2402445364924636,
      "learning_rate": 5.4007016058211035e-05,
      "loss": 1.0375,
      "step": 9192
    },
    {
      "epoch": 0.6626540762632451,
      "grad_norm": 0.13383121814409002,
      "learning_rate": 5.3986286086290746e-05,
      "loss": 1.0277,
      "step": 9193
    },
    {
      "epoch": 0.662726158725582,
      "grad_norm": 0.12351518623986993,
      "learning_rate": 5.396555862252952e-05,
      "loss": 1.0527,
      "step": 9194
    },
    {
      "epoch": 0.662798241187919,
      "grad_norm": 0.12511193889028724,
      "learning_rate": 5.3944833668057185e-05,
      "loss": 1.0689,
      "step": 9195
    },
    {
      "epoch": 0.6628703236502559,
      "grad_norm": 0.13232782364923146,
      "learning_rate": 5.3924111224003424e-05,
      "loss": 1.0105,
      "step": 9196
    },
    {
      "epoch": 0.6629424061125928,
      "grad_norm": 0.11857061838902162,
      "learning_rate": 5.3903391291497815e-05,
      "loss": 1.1085,
      "step": 9197
    },
    {
      "epoch": 0.6630144885749297,
      "grad_norm": 0.12909949258587008,
      "learning_rate": 5.3882673871669766e-05,
      "loss": 1.0019,
      "step": 9198
    },
    {
      "epoch": 0.6630865710372666,
      "grad_norm": 0.41844515343877287,
      "learning_rate": 5.386195896564862e-05,
      "loss": 1.0641,
      "step": 9199
    },
    {
      "epoch": 0.6631586534996036,
      "grad_norm": 0.12720273526703485,
      "learning_rate": 5.384124657456339e-05,
      "loss": 1.0035,
      "step": 9200
    },
    {
      "epoch": 0.6632307359619405,
      "grad_norm": 0.15131180368363487,
      "learning_rate": 5.382053669954321e-05,
      "loss": 1.0005,
      "step": 9201
    },
    {
      "epoch": 0.6633028184242774,
      "grad_norm": 0.1548278057727988,
      "learning_rate": 5.37998293417169e-05,
      "loss": 1.0121,
      "step": 9202
    },
    {
      "epoch": 0.6633749008866143,
      "grad_norm": 0.13005678278639857,
      "learning_rate": 5.377912450221326e-05,
      "loss": 1.0011,
      "step": 9203
    },
    {
      "epoch": 0.6634469833489512,
      "grad_norm": 0.1263027659549693,
      "learning_rate": 5.375842218216076e-05,
      "loss": 1.0183,
      "step": 9204
    },
    {
      "epoch": 0.6635190658112882,
      "grad_norm": 0.12207882310120896,
      "learning_rate": 5.373772238268798e-05,
      "loss": 1.0652,
      "step": 9205
    },
    {
      "epoch": 0.663591148273625,
      "grad_norm": 0.11560153810116201,
      "learning_rate": 5.37170251049232e-05,
      "loss": 1.0577,
      "step": 9206
    },
    {
      "epoch": 0.6636632307359619,
      "grad_norm": 0.13827055114515283,
      "learning_rate": 5.369633034999466e-05,
      "loss": 0.9094,
      "step": 9207
    },
    {
      "epoch": 0.6637353131982988,
      "grad_norm": 0.11092314646185877,
      "learning_rate": 5.367563811903027e-05,
      "loss": 1.0089,
      "step": 9208
    },
    {
      "epoch": 0.6638073956606357,
      "grad_norm": 0.17495703952394343,
      "learning_rate": 5.365494841315808e-05,
      "loss": 1.0229,
      "step": 9209
    },
    {
      "epoch": 0.6638794781229727,
      "grad_norm": 0.11789811297973377,
      "learning_rate": 5.363426123350586e-05,
      "loss": 1.0903,
      "step": 9210
    },
    {
      "epoch": 0.6639515605853096,
      "grad_norm": 0.12435275638158383,
      "learning_rate": 5.361357658120114e-05,
      "loss": 1.0662,
      "step": 9211
    },
    {
      "epoch": 0.6640236430476465,
      "grad_norm": 0.11906038792005502,
      "learning_rate": 5.35928944573715e-05,
      "loss": 1.0991,
      "step": 9212
    },
    {
      "epoch": 0.6640957255099834,
      "grad_norm": 0.12226852303182419,
      "learning_rate": 5.3572214863144246e-05,
      "loss": 0.9843,
      "step": 9213
    },
    {
      "epoch": 0.6641678079723203,
      "grad_norm": 0.12145494830325104,
      "learning_rate": 5.3551537799646704e-05,
      "loss": 1.0429,
      "step": 9214
    },
    {
      "epoch": 0.6642398904346573,
      "grad_norm": 0.1406749493641633,
      "learning_rate": 5.353086326800586e-05,
      "loss": 1.0464,
      "step": 9215
    },
    {
      "epoch": 0.6643119728969942,
      "grad_norm": 0.13402087004548224,
      "learning_rate": 5.351019126934869e-05,
      "loss": 1.025,
      "step": 9216
    },
    {
      "epoch": 0.6643840553593311,
      "grad_norm": 0.11328767714467045,
      "learning_rate": 5.348952180480196e-05,
      "loss": 1.0017,
      "step": 9217
    },
    {
      "epoch": 0.664456137821668,
      "grad_norm": 0.10996154517525457,
      "learning_rate": 5.346885487549247e-05,
      "loss": 1.0278,
      "step": 9218
    },
    {
      "epoch": 0.6645282202840049,
      "grad_norm": 0.13746775643659695,
      "learning_rate": 5.3448190482546634e-05,
      "loss": 1.1088,
      "step": 9219
    },
    {
      "epoch": 0.6646003027463419,
      "grad_norm": 0.1879360277898876,
      "learning_rate": 5.342752862709088e-05,
      "loss": 1.0047,
      "step": 9220
    },
    {
      "epoch": 0.6646723852086788,
      "grad_norm": 0.1415261603909499,
      "learning_rate": 5.340686931025146e-05,
      "loss": 1.0461,
      "step": 9221
    },
    {
      "epoch": 0.6647444676710157,
      "grad_norm": 0.14255086787358523,
      "learning_rate": 5.33862125331545e-05,
      "loss": 0.9827,
      "step": 9222
    },
    {
      "epoch": 0.6648165501333525,
      "grad_norm": 0.11194955245679086,
      "learning_rate": 5.336555829692598e-05,
      "loss": 1.0166,
      "step": 9223
    },
    {
      "epoch": 0.6648886325956894,
      "grad_norm": 0.13532782480925554,
      "learning_rate": 5.3344906602691726e-05,
      "loss": 0.9882,
      "step": 9224
    },
    {
      "epoch": 0.6649607150580263,
      "grad_norm": 0.12770690485788788,
      "learning_rate": 5.332425745157745e-05,
      "loss": 1.0346,
      "step": 9225
    },
    {
      "epoch": 0.6650327975203633,
      "grad_norm": 0.45763860865793066,
      "learning_rate": 5.3303610844708726e-05,
      "loss": 1.0814,
      "step": 9226
    },
    {
      "epoch": 0.6651048799827002,
      "grad_norm": 0.12090828940307574,
      "learning_rate": 5.3282966783210964e-05,
      "loss": 1.0752,
      "step": 9227
    },
    {
      "epoch": 0.6651769624450371,
      "grad_norm": 0.11845819504775995,
      "learning_rate": 5.3262325268209454e-05,
      "loss": 1.0517,
      "step": 9228
    },
    {
      "epoch": 0.665249044907374,
      "grad_norm": 0.14079557821211874,
      "learning_rate": 5.324168630082935e-05,
      "loss": 1.0093,
      "step": 9229
    },
    {
      "epoch": 0.6653211273697109,
      "grad_norm": 0.13226720740712933,
      "learning_rate": 5.322104988219565e-05,
      "loss": 1.0147,
      "step": 9230
    },
    {
      "epoch": 0.6653932098320479,
      "grad_norm": 0.1433297053741452,
      "learning_rate": 5.320041601343322e-05,
      "loss": 1.0309,
      "step": 9231
    },
    {
      "epoch": 0.6654652922943848,
      "grad_norm": 0.17569840560399136,
      "learning_rate": 5.317978469566681e-05,
      "loss": 1.0417,
      "step": 9232
    },
    {
      "epoch": 0.6655373747567217,
      "grad_norm": 0.15604251885550804,
      "learning_rate": 5.3159155930021e-05,
      "loss": 0.9755,
      "step": 9233
    },
    {
      "epoch": 0.6656094572190586,
      "grad_norm": 0.13734642931055585,
      "learning_rate": 5.313852971762023e-05,
      "loss": 1.053,
      "step": 9234
    },
    {
      "epoch": 0.6656815396813955,
      "grad_norm": 0.11537481712442324,
      "learning_rate": 5.311790605958884e-05,
      "loss": 1.1243,
      "step": 9235
    },
    {
      "epoch": 0.6657536221437325,
      "grad_norm": 0.12345694251812275,
      "learning_rate": 5.3097284957050974e-05,
      "loss": 1.1016,
      "step": 9236
    },
    {
      "epoch": 0.6658257046060694,
      "grad_norm": 0.12342403799794203,
      "learning_rate": 5.307666641113069e-05,
      "loss": 1.0374,
      "step": 9237
    },
    {
      "epoch": 0.6658977870684063,
      "grad_norm": 0.16220494401615446,
      "learning_rate": 5.3056050422951874e-05,
      "loss": 1.1709,
      "step": 9238
    },
    {
      "epoch": 0.6659698695307432,
      "grad_norm": 0.36169105131314017,
      "learning_rate": 5.3035436993638286e-05,
      "loss": 1.0422,
      "step": 9239
    },
    {
      "epoch": 0.6660419519930801,
      "grad_norm": 0.14882906612291535,
      "learning_rate": 5.301482612431354e-05,
      "loss": 1.0003,
      "step": 9240
    },
    {
      "epoch": 0.666114034455417,
      "grad_norm": 0.1533665185530863,
      "learning_rate": 5.29942178161011e-05,
      "loss": 1.0216,
      "step": 9241
    },
    {
      "epoch": 0.6661861169177539,
      "grad_norm": 0.14808096615119473,
      "learning_rate": 5.297361207012432e-05,
      "loss": 1.0293,
      "step": 9242
    },
    {
      "epoch": 0.6662581993800908,
      "grad_norm": 0.12147808462963143,
      "learning_rate": 5.295300888750644e-05,
      "loss": 1.0507,
      "step": 9243
    },
    {
      "epoch": 0.6663302818424277,
      "grad_norm": 0.12816870306560352,
      "learning_rate": 5.293240826937038e-05,
      "loss": 1.0575,
      "step": 9244
    },
    {
      "epoch": 0.6664023643047646,
      "grad_norm": 0.11500849542611058,
      "learning_rate": 5.29118102168392e-05,
      "loss": 1.0088,
      "step": 9245
    },
    {
      "epoch": 0.6664744467671015,
      "grad_norm": 0.12246352535814041,
      "learning_rate": 5.289121473103562e-05,
      "loss": 0.9957,
      "step": 9246
    },
    {
      "epoch": 0.6665465292294385,
      "grad_norm": 0.7457782164141348,
      "learning_rate": 5.287062181308231e-05,
      "loss": 1.0359,
      "step": 9247
    },
    {
      "epoch": 0.6666186116917754,
      "grad_norm": 0.13050332187059052,
      "learning_rate": 5.285003146410168e-05,
      "loss": 1.0257,
      "step": 9248
    },
    {
      "epoch": 0.6666906941541123,
      "grad_norm": 0.1259492512137833,
      "learning_rate": 5.282944368521618e-05,
      "loss": 1.0048,
      "step": 9249
    },
    {
      "epoch": 0.6667627766164492,
      "grad_norm": 0.12278389846255876,
      "learning_rate": 5.2808858477548014e-05,
      "loss": 1.0017,
      "step": 9250
    },
    {
      "epoch": 0.6668348590787861,
      "grad_norm": 0.12916170719763223,
      "learning_rate": 5.278827584221928e-05,
      "loss": 1.0498,
      "step": 9251
    },
    {
      "epoch": 0.6669069415411231,
      "grad_norm": 0.11159890989139025,
      "learning_rate": 5.276769578035179e-05,
      "loss": 1.082,
      "step": 9252
    },
    {
      "epoch": 0.66697902400346,
      "grad_norm": 0.11664991309036679,
      "learning_rate": 5.274711829306749e-05,
      "loss": 1.0526,
      "step": 9253
    },
    {
      "epoch": 0.6670511064657969,
      "grad_norm": 0.1280750049918235,
      "learning_rate": 5.272654338148796e-05,
      "loss": 0.9948,
      "step": 9254
    },
    {
      "epoch": 0.6671231889281338,
      "grad_norm": 0.11576521841984758,
      "learning_rate": 5.270597104673478e-05,
      "loss": 0.9946,
      "step": 9255
    },
    {
      "epoch": 0.6671952713904707,
      "grad_norm": 0.11919464887406506,
      "learning_rate": 5.2685401289929245e-05,
      "loss": 1.0486,
      "step": 9256
    },
    {
      "epoch": 0.6672673538528077,
      "grad_norm": 0.10903913195895301,
      "learning_rate": 5.2664834112192594e-05,
      "loss": 1.0573,
      "step": 9257
    },
    {
      "epoch": 0.6673394363151445,
      "grad_norm": 0.1263276637361464,
      "learning_rate": 5.264426951464599e-05,
      "loss": 1.0087,
      "step": 9258
    },
    {
      "epoch": 0.6674115187774814,
      "grad_norm": 0.14505508158909725,
      "learning_rate": 5.26237074984104e-05,
      "loss": 1.053,
      "step": 9259
    },
    {
      "epoch": 0.6674836012398183,
      "grad_norm": 0.12605908273799565,
      "learning_rate": 5.260314806460655e-05,
      "loss": 0.9769,
      "step": 9260
    },
    {
      "epoch": 0.6675556837021552,
      "grad_norm": 0.12806807698242526,
      "learning_rate": 5.2582591214355113e-05,
      "loss": 1.0569,
      "step": 9261
    },
    {
      "epoch": 0.6676277661644922,
      "grad_norm": 0.14908072473348896,
      "learning_rate": 5.25620369487767e-05,
      "loss": 1.0587,
      "step": 9262
    },
    {
      "epoch": 0.6676998486268291,
      "grad_norm": 0.13084762772145128,
      "learning_rate": 5.254148526899171e-05,
      "loss": 1.0456,
      "step": 9263
    },
    {
      "epoch": 0.667771931089166,
      "grad_norm": 0.11424423485734875,
      "learning_rate": 5.252093617612031e-05,
      "loss": 1.0665,
      "step": 9264
    },
    {
      "epoch": 0.6678440135515029,
      "grad_norm": 0.15337972689674403,
      "learning_rate": 5.250038967128264e-05,
      "loss": 1.0153,
      "step": 9265
    },
    {
      "epoch": 0.6679160960138398,
      "grad_norm": 0.1348667850442683,
      "learning_rate": 5.2479845755598656e-05,
      "loss": 1.0166,
      "step": 9266
    },
    {
      "epoch": 0.6679881784761768,
      "grad_norm": 0.11661878923355855,
      "learning_rate": 5.245930443018826e-05,
      "loss": 1.0473,
      "step": 9267
    },
    {
      "epoch": 0.6680602609385137,
      "grad_norm": 0.14839695286929513,
      "learning_rate": 5.2438765696171054e-05,
      "loss": 1.0622,
      "step": 9268
    },
    {
      "epoch": 0.6681323434008506,
      "grad_norm": 0.1238285728354813,
      "learning_rate": 5.241822955466662e-05,
      "loss": 1.1023,
      "step": 9269
    },
    {
      "epoch": 0.6682044258631875,
      "grad_norm": 0.13326957589830793,
      "learning_rate": 5.2397696006794306e-05,
      "loss": 1.0334,
      "step": 9270
    },
    {
      "epoch": 0.6682765083255244,
      "grad_norm": 0.11492271444039596,
      "learning_rate": 5.2377165053673505e-05,
      "loss": 1.0473,
      "step": 9271
    },
    {
      "epoch": 0.6683485907878614,
      "grad_norm": 0.13238636062888878,
      "learning_rate": 5.235663669642321e-05,
      "loss": 1.0064,
      "step": 9272
    },
    {
      "epoch": 0.6684206732501983,
      "grad_norm": 0.134263047432655,
      "learning_rate": 5.233611093616245e-05,
      "loss": 1.0258,
      "step": 9273
    },
    {
      "epoch": 0.6684927557125352,
      "grad_norm": 0.12193703360932047,
      "learning_rate": 5.231558777401002e-05,
      "loss": 1.0306,
      "step": 9274
    },
    {
      "epoch": 0.668564838174872,
      "grad_norm": 0.12185986187029099,
      "learning_rate": 5.229506721108473e-05,
      "loss": 1.0269,
      "step": 9275
    },
    {
      "epoch": 0.6686369206372089,
      "grad_norm": 0.1445722451303639,
      "learning_rate": 5.227454924850501e-05,
      "loss": 0.9897,
      "step": 9276
    },
    {
      "epoch": 0.6687090030995458,
      "grad_norm": 0.22923455432358208,
      "learning_rate": 5.225403388738933e-05,
      "loss": 1.0902,
      "step": 9277
    },
    {
      "epoch": 0.6687810855618828,
      "grad_norm": 0.19435125784498175,
      "learning_rate": 5.2233521128855956e-05,
      "loss": 1.0305,
      "step": 9278
    },
    {
      "epoch": 0.6688531680242197,
      "grad_norm": 0.5602189126983482,
      "learning_rate": 5.2213010974023004e-05,
      "loss": 1.0261,
      "step": 9279
    },
    {
      "epoch": 0.6689252504865566,
      "grad_norm": 0.13111030405502477,
      "learning_rate": 5.219250342400845e-05,
      "loss": 1.053,
      "step": 9280
    },
    {
      "epoch": 0.6689973329488935,
      "grad_norm": 0.11602281579676618,
      "learning_rate": 5.2171998479930185e-05,
      "loss": 1.0315,
      "step": 9281
    },
    {
      "epoch": 0.6690694154112304,
      "grad_norm": 0.13540241686863064,
      "learning_rate": 5.2151496142905876e-05,
      "loss": 1.0713,
      "step": 9282
    },
    {
      "epoch": 0.6691414978735674,
      "grad_norm": 0.1171377392766916,
      "learning_rate": 5.213099641405309e-05,
      "loss": 1.1032,
      "step": 9283
    },
    {
      "epoch": 0.6692135803359043,
      "grad_norm": 0.10109919770562373,
      "learning_rate": 5.211049929448926e-05,
      "loss": 0.9902,
      "step": 9284
    },
    {
      "epoch": 0.6692856627982412,
      "grad_norm": 0.11003183980964293,
      "learning_rate": 5.209000478533163e-05,
      "loss": 1.0131,
      "step": 9285
    },
    {
      "epoch": 0.6693577452605781,
      "grad_norm": 0.1369530968328839,
      "learning_rate": 5.206951288769736e-05,
      "loss": 0.9819,
      "step": 9286
    },
    {
      "epoch": 0.669429827722915,
      "grad_norm": 0.16833680772874218,
      "learning_rate": 5.204902360270344e-05,
      "loss": 0.9982,
      "step": 9287
    },
    {
      "epoch": 0.669501910185252,
      "grad_norm": 0.12227985198428218,
      "learning_rate": 5.202853693146672e-05,
      "loss": 1.024,
      "step": 9288
    },
    {
      "epoch": 0.6695739926475889,
      "grad_norm": 0.1064091337896857,
      "learning_rate": 5.2008052875103886e-05,
      "loss": 1.0571,
      "step": 9289
    },
    {
      "epoch": 0.6696460751099258,
      "grad_norm": 0.13032142384753556,
      "learning_rate": 5.198757143473154e-05,
      "loss": 1.1042,
      "step": 9290
    },
    {
      "epoch": 0.6697181575722627,
      "grad_norm": 0.13019353539409442,
      "learning_rate": 5.196709261146606e-05,
      "loss": 1.0291,
      "step": 9291
    },
    {
      "epoch": 0.6697902400345995,
      "grad_norm": 0.11203472660631603,
      "learning_rate": 5.194661640642377e-05,
      "loss": 1.0308,
      "step": 9292
    },
    {
      "epoch": 0.6698623224969364,
      "grad_norm": 0.13548470070713392,
      "learning_rate": 5.192614282072077e-05,
      "loss": 1.0524,
      "step": 9293
    },
    {
      "epoch": 0.6699344049592734,
      "grad_norm": 0.13474019861548864,
      "learning_rate": 5.190567185547306e-05,
      "loss": 1.1135,
      "step": 9294
    },
    {
      "epoch": 0.6700064874216103,
      "grad_norm": 0.22494442941236456,
      "learning_rate": 5.188520351179651e-05,
      "loss": 1.0576,
      "step": 9295
    },
    {
      "epoch": 0.6700785698839472,
      "grad_norm": 0.23718591634991978,
      "learning_rate": 5.186473779080682e-05,
      "loss": 1.0566,
      "step": 9296
    },
    {
      "epoch": 0.6701506523462841,
      "grad_norm": 0.10616163676164224,
      "learning_rate": 5.184427469361954e-05,
      "loss": 0.9573,
      "step": 9297
    },
    {
      "epoch": 0.670222734808621,
      "grad_norm": 0.11639863913818538,
      "learning_rate": 5.182381422135011e-05,
      "loss": 0.996,
      "step": 9298
    },
    {
      "epoch": 0.670294817270958,
      "grad_norm": 0.13749799768755006,
      "learning_rate": 5.18033563751138e-05,
      "loss": 1.0866,
      "step": 9299
    },
    {
      "epoch": 0.6703668997332949,
      "grad_norm": 0.12258479757659348,
      "learning_rate": 5.178290115602578e-05,
      "loss": 1.0877,
      "step": 9300
    },
    {
      "epoch": 0.6704389821956318,
      "grad_norm": 0.11799831027611826,
      "learning_rate": 5.176244856520094e-05,
      "loss": 0.9835,
      "step": 9301
    },
    {
      "epoch": 0.6705110646579687,
      "grad_norm": 0.12109020789062414,
      "learning_rate": 5.174199860375424e-05,
      "loss": 1.0187,
      "step": 9302
    },
    {
      "epoch": 0.6705831471203056,
      "grad_norm": 0.20255730893208054,
      "learning_rate": 5.1721551272800326e-05,
      "loss": 1.0878,
      "step": 9303
    },
    {
      "epoch": 0.6706552295826426,
      "grad_norm": 0.1132392403187892,
      "learning_rate": 5.1701106573453836e-05,
      "loss": 1.1264,
      "step": 9304
    },
    {
      "epoch": 0.6707273120449795,
      "grad_norm": 0.15504279899249482,
      "learning_rate": 5.1680664506829045e-05,
      "loss": 1.063,
      "step": 9305
    },
    {
      "epoch": 0.6707993945073164,
      "grad_norm": 0.10743493867462658,
      "learning_rate": 5.1660225074040355e-05,
      "loss": 0.9732,
      "step": 9306
    },
    {
      "epoch": 0.6708714769696533,
      "grad_norm": 0.12542653612633448,
      "learning_rate": 5.163978827620186e-05,
      "loss": 1.122,
      "step": 9307
    },
    {
      "epoch": 0.6709435594319902,
      "grad_norm": 0.18385620892347718,
      "learning_rate": 5.1619354114427595e-05,
      "loss": 1.0301,
      "step": 9308
    },
    {
      "epoch": 0.6710156418943272,
      "grad_norm": 0.14854261339299976,
      "learning_rate": 5.15989225898313e-05,
      "loss": 1.0647,
      "step": 9309
    },
    {
      "epoch": 0.671087724356664,
      "grad_norm": 0.1513858098230488,
      "learning_rate": 5.157849370352671e-05,
      "loss": 1.067,
      "step": 9310
    },
    {
      "epoch": 0.6711598068190009,
      "grad_norm": 0.13137255567293124,
      "learning_rate": 5.155806745662743e-05,
      "loss": 1.0685,
      "step": 9311
    },
    {
      "epoch": 0.6712318892813378,
      "grad_norm": 0.10801519860133771,
      "learning_rate": 5.15376438502469e-05,
      "loss": 1.0104,
      "step": 9312
    },
    {
      "epoch": 0.6713039717436747,
      "grad_norm": 0.16167322697256536,
      "learning_rate": 5.151722288549828e-05,
      "loss": 0.9763,
      "step": 9313
    },
    {
      "epoch": 0.6713760542060117,
      "grad_norm": 0.11142651326978593,
      "learning_rate": 5.149680456349473e-05,
      "loss": 1.0272,
      "step": 9314
    },
    {
      "epoch": 0.6714481366683486,
      "grad_norm": 0.11775707039163685,
      "learning_rate": 5.147638888534929e-05,
      "loss": 1.0811,
      "step": 9315
    },
    {
      "epoch": 0.6715202191306855,
      "grad_norm": 0.12906695244895794,
      "learning_rate": 5.1455975852174784e-05,
      "loss": 1.0233,
      "step": 9316
    },
    {
      "epoch": 0.6715923015930224,
      "grad_norm": 0.13418490000540273,
      "learning_rate": 5.143556546508386e-05,
      "loss": 1.0176,
      "step": 9317
    },
    {
      "epoch": 0.6716643840553593,
      "grad_norm": 0.11380563620764898,
      "learning_rate": 5.141515772518903e-05,
      "loss": 1.0133,
      "step": 9318
    },
    {
      "epoch": 0.6717364665176963,
      "grad_norm": 0.12018601492851484,
      "learning_rate": 5.1394752633602816e-05,
      "loss": 1.0466,
      "step": 9319
    },
    {
      "epoch": 0.6718085489800332,
      "grad_norm": 0.11414191715718897,
      "learning_rate": 5.1374350191437446e-05,
      "loss": 1.0388,
      "step": 9320
    },
    {
      "epoch": 0.6718806314423701,
      "grad_norm": 0.12540233777183124,
      "learning_rate": 5.135395039980497e-05,
      "loss": 0.9826,
      "step": 9321
    },
    {
      "epoch": 0.671952713904707,
      "grad_norm": 0.11137294787182794,
      "learning_rate": 5.133355325981741e-05,
      "loss": 1.0478,
      "step": 9322
    },
    {
      "epoch": 0.6720247963670439,
      "grad_norm": 0.12294321915789774,
      "learning_rate": 5.131315877258653e-05,
      "loss": 1.0242,
      "step": 9323
    },
    {
      "epoch": 0.6720968788293809,
      "grad_norm": 0.11757258250755585,
      "learning_rate": 5.1292766939224133e-05,
      "loss": 1.0371,
      "step": 9324
    },
    {
      "epoch": 0.6721689612917178,
      "grad_norm": 0.12093641206260715,
      "learning_rate": 5.1272377760841664e-05,
      "loss": 1.0158,
      "step": 9325
    },
    {
      "epoch": 0.6722410437540547,
      "grad_norm": 0.18589009988314395,
      "learning_rate": 5.125199123855051e-05,
      "loss": 1.0969,
      "step": 9326
    },
    {
      "epoch": 0.6723131262163915,
      "grad_norm": 0.11487774187320957,
      "learning_rate": 5.1231607373461934e-05,
      "loss": 1.041,
      "step": 9327
    },
    {
      "epoch": 0.6723852086787284,
      "grad_norm": 0.11833018510905642,
      "learning_rate": 5.121122616668712e-05,
      "loss": 1.052,
      "step": 9328
    },
    {
      "epoch": 0.6724572911410653,
      "grad_norm": 0.14432632538064072,
      "learning_rate": 5.1190847619336904e-05,
      "loss": 1.0581,
      "step": 9329
    },
    {
      "epoch": 0.6725293736034023,
      "grad_norm": 0.11151971416116775,
      "learning_rate": 5.1170471732522165e-05,
      "loss": 1.0188,
      "step": 9330
    },
    {
      "epoch": 0.6726014560657392,
      "grad_norm": 0.113730789526338,
      "learning_rate": 5.115009850735355e-05,
      "loss": 1.0359,
      "step": 9331
    },
    {
      "epoch": 0.6726735385280761,
      "grad_norm": 0.12692632112614785,
      "learning_rate": 5.1129727944941594e-05,
      "loss": 1.0375,
      "step": 9332
    },
    {
      "epoch": 0.672745620990413,
      "grad_norm": 0.13485774304715759,
      "learning_rate": 5.110936004639668e-05,
      "loss": 1.0132,
      "step": 9333
    },
    {
      "epoch": 0.6728177034527499,
      "grad_norm": 0.11020800715820704,
      "learning_rate": 5.108899481282903e-05,
      "loss": 1.0358,
      "step": 9334
    },
    {
      "epoch": 0.6728897859150869,
      "grad_norm": 0.12308241202636641,
      "learning_rate": 5.1068632245348726e-05,
      "loss": 1.0575,
      "step": 9335
    },
    {
      "epoch": 0.6729618683774238,
      "grad_norm": 0.11960844628930786,
      "learning_rate": 5.104827234506572e-05,
      "loss": 0.9573,
      "step": 9336
    },
    {
      "epoch": 0.6730339508397607,
      "grad_norm": 0.10443593089249859,
      "learning_rate": 5.1027915113089796e-05,
      "loss": 0.9958,
      "step": 9337
    },
    {
      "epoch": 0.6731060333020976,
      "grad_norm": 0.11664558611153628,
      "learning_rate": 5.100756055053063e-05,
      "loss": 1.0073,
      "step": 9338
    },
    {
      "epoch": 0.6731781157644345,
      "grad_norm": 0.25035947381791307,
      "learning_rate": 5.098720865849771e-05,
      "loss": 1.16,
      "step": 9339
    },
    {
      "epoch": 0.6732501982267715,
      "grad_norm": 0.11011588477723011,
      "learning_rate": 5.096685943810039e-05,
      "loss": 1.108,
      "step": 9340
    },
    {
      "epoch": 0.6733222806891084,
      "grad_norm": 0.1493154069791278,
      "learning_rate": 5.0946512890447896e-05,
      "loss": 1.0486,
      "step": 9341
    },
    {
      "epoch": 0.6733943631514453,
      "grad_norm": 0.12554573040508948,
      "learning_rate": 5.09261690166493e-05,
      "loss": 1.1164,
      "step": 9342
    },
    {
      "epoch": 0.6734664456137822,
      "grad_norm": 0.29505788156943047,
      "learning_rate": 5.0905827817813526e-05,
      "loss": 0.9977,
      "step": 9343
    },
    {
      "epoch": 0.673538528076119,
      "grad_norm": 0.1363039946936859,
      "learning_rate": 5.088548929504933e-05,
      "loss": 1.0371,
      "step": 9344
    },
    {
      "epoch": 0.673610610538456,
      "grad_norm": 0.11214543947279854,
      "learning_rate": 5.086515344946536e-05,
      "loss": 1.0595,
      "step": 9345
    },
    {
      "epoch": 0.6736826930007929,
      "grad_norm": 0.1349821949093559,
      "learning_rate": 5.08448202821701e-05,
      "loss": 1.0292,
      "step": 9346
    },
    {
      "epoch": 0.6737547754631298,
      "grad_norm": 0.12307197546429259,
      "learning_rate": 5.08244897942719e-05,
      "loss": 1.0722,
      "step": 9347
    },
    {
      "epoch": 0.6738268579254667,
      "grad_norm": 0.10737115296103954,
      "learning_rate": 5.080416198687894e-05,
      "loss": 1.0231,
      "step": 9348
    },
    {
      "epoch": 0.6738989403878036,
      "grad_norm": 0.11613514388051603,
      "learning_rate": 5.078383686109926e-05,
      "loss": 1.1198,
      "step": 9349
    },
    {
      "epoch": 0.6739710228501405,
      "grad_norm": 0.12033340371263164,
      "learning_rate": 5.076351441804079e-05,
      "loss": 1.0093,
      "step": 9350
    },
    {
      "epoch": 0.6740431053124775,
      "grad_norm": 0.12935478895163152,
      "learning_rate": 5.074319465881126e-05,
      "loss": 0.9614,
      "step": 9351
    },
    {
      "epoch": 0.6741151877748144,
      "grad_norm": 0.13072000172003403,
      "learning_rate": 5.0722877584518266e-05,
      "loss": 1.0281,
      "step": 9352
    },
    {
      "epoch": 0.6741872702371513,
      "grad_norm": 0.10640205082547967,
      "learning_rate": 5.070256319626935e-05,
      "loss": 1.0777,
      "step": 9353
    },
    {
      "epoch": 0.6742593526994882,
      "grad_norm": 0.12038581322465075,
      "learning_rate": 5.068225149517167e-05,
      "loss": 1.0405,
      "step": 9354
    },
    {
      "epoch": 0.6743314351618251,
      "grad_norm": 0.1542622705013238,
      "learning_rate": 5.0661942482332556e-05,
      "loss": 1.06,
      "step": 9355
    },
    {
      "epoch": 0.6744035176241621,
      "grad_norm": 0.12289870268869564,
      "learning_rate": 5.0641636158858954e-05,
      "loss": 1.11,
      "step": 9356
    },
    {
      "epoch": 0.674475600086499,
      "grad_norm": 0.12039798564986638,
      "learning_rate": 5.062133252585779e-05,
      "loss": 0.9917,
      "step": 9357
    },
    {
      "epoch": 0.6745476825488359,
      "grad_norm": 0.12112571944757866,
      "learning_rate": 5.0601031584435675e-05,
      "loss": 1.0117,
      "step": 9358
    },
    {
      "epoch": 0.6746197650111728,
      "grad_norm": 0.11768670188517921,
      "learning_rate": 5.058073333569933e-05,
      "loss": 1.0734,
      "step": 9359
    },
    {
      "epoch": 0.6746918474735097,
      "grad_norm": 0.1265288547070817,
      "learning_rate": 5.056043778075512e-05,
      "loss": 1.0483,
      "step": 9360
    },
    {
      "epoch": 0.6747639299358467,
      "grad_norm": 0.1130677549258569,
      "learning_rate": 5.05401449207094e-05,
      "loss": 0.9967,
      "step": 9361
    },
    {
      "epoch": 0.6748360123981835,
      "grad_norm": 0.10965807179706415,
      "learning_rate": 5.0519854756668184e-05,
      "loss": 1.059,
      "step": 9362
    },
    {
      "epoch": 0.6749080948605204,
      "grad_norm": 0.13033405621815586,
      "learning_rate": 5.0499567289737584e-05,
      "loss": 0.9971,
      "step": 9363
    },
    {
      "epoch": 0.6749801773228573,
      "grad_norm": 0.10832620961513963,
      "learning_rate": 5.047928252102345e-05,
      "loss": 1.1024,
      "step": 9364
    },
    {
      "epoch": 0.6750522597851942,
      "grad_norm": 0.1501610749838149,
      "learning_rate": 5.04590004516314e-05,
      "loss": 1.0483,
      "step": 9365
    },
    {
      "epoch": 0.6751243422475312,
      "grad_norm": 0.10900055758618755,
      "learning_rate": 5.043872108266703e-05,
      "loss": 1.0092,
      "step": 9366
    },
    {
      "epoch": 0.6751964247098681,
      "grad_norm": 0.125894841096761,
      "learning_rate": 5.041844441523571e-05,
      "loss": 1.1008,
      "step": 9367
    },
    {
      "epoch": 0.675268507172205,
      "grad_norm": 0.18115073920486635,
      "learning_rate": 5.039817045044282e-05,
      "loss": 0.9035,
      "step": 9368
    },
    {
      "epoch": 0.6753405896345419,
      "grad_norm": 0.11721341778498985,
      "learning_rate": 5.037789918939335e-05,
      "loss": 0.9686,
      "step": 9369
    },
    {
      "epoch": 0.6754126720968788,
      "grad_norm": 0.12398324105609475,
      "learning_rate": 5.035763063319232e-05,
      "loss": 1.081,
      "step": 9370
    },
    {
      "epoch": 0.6754847545592157,
      "grad_norm": 0.10591972777698756,
      "learning_rate": 5.0337364782944475e-05,
      "loss": 1.0479,
      "step": 9371
    },
    {
      "epoch": 0.6755568370215527,
      "grad_norm": 0.10870093413906567,
      "learning_rate": 5.031710163975463e-05,
      "loss": 0.9737,
      "step": 9372
    },
    {
      "epoch": 0.6756289194838896,
      "grad_norm": 0.11845878769500762,
      "learning_rate": 5.029684120472719e-05,
      "loss": 1.0136,
      "step": 9373
    },
    {
      "epoch": 0.6757010019462265,
      "grad_norm": 0.1318527721997842,
      "learning_rate": 5.0276583478966556e-05,
      "loss": 1.0983,
      "step": 9374
    },
    {
      "epoch": 0.6757730844085634,
      "grad_norm": 0.1122414353864023,
      "learning_rate": 5.025632846357695e-05,
      "loss": 1.0289,
      "step": 9375
    },
    {
      "epoch": 0.6758451668709003,
      "grad_norm": 0.12473958527170995,
      "learning_rate": 5.023607615966247e-05,
      "loss": 1.0737,
      "step": 9376
    },
    {
      "epoch": 0.6759172493332373,
      "grad_norm": 0.1307882808802339,
      "learning_rate": 5.021582656832704e-05,
      "loss": 0.9363,
      "step": 9377
    },
    {
      "epoch": 0.6759893317955742,
      "grad_norm": 0.13834125898708785,
      "learning_rate": 5.0195579690674447e-05,
      "loss": 1.1002,
      "step": 9378
    },
    {
      "epoch": 0.676061414257911,
      "grad_norm": 0.12925053186228938,
      "learning_rate": 5.017533552780832e-05,
      "loss": 1.0424,
      "step": 9379
    },
    {
      "epoch": 0.6761334967202479,
      "grad_norm": 0.11940999588507778,
      "learning_rate": 5.015509408083215e-05,
      "loss": 1.0709,
      "step": 9380
    },
    {
      "epoch": 0.6762055791825848,
      "grad_norm": 0.11684041229045956,
      "learning_rate": 5.0134855350849284e-05,
      "loss": 1.0203,
      "step": 9381
    },
    {
      "epoch": 0.6762776616449218,
      "grad_norm": 0.12336749211140197,
      "learning_rate": 5.011461933896291e-05,
      "loss": 1.0786,
      "step": 9382
    },
    {
      "epoch": 0.6763497441072587,
      "grad_norm": 0.13300085327859942,
      "learning_rate": 5.0094386046276054e-05,
      "loss": 1.0747,
      "step": 9383
    },
    {
      "epoch": 0.6764218265695956,
      "grad_norm": 0.12182618817876424,
      "learning_rate": 5.007415547389165e-05,
      "loss": 1.0549,
      "step": 9384
    },
    {
      "epoch": 0.6764939090319325,
      "grad_norm": 0.11563512780594083,
      "learning_rate": 5.00539276229124e-05,
      "loss": 0.9839,
      "step": 9385
    },
    {
      "epoch": 0.6765659914942694,
      "grad_norm": 0.7277327487048859,
      "learning_rate": 5.003370249444093e-05,
      "loss": 1.0036,
      "step": 9386
    },
    {
      "epoch": 0.6766380739566064,
      "grad_norm": 0.14583727593851603,
      "learning_rate": 5.001348008957969e-05,
      "loss": 0.9897,
      "step": 9387
    },
    {
      "epoch": 0.6767101564189433,
      "grad_norm": 0.13750015859890038,
      "learning_rate": 4.999326040943098e-05,
      "loss": 1.1393,
      "step": 9388
    },
    {
      "epoch": 0.6767822388812802,
      "grad_norm": 0.13973677262875828,
      "learning_rate": 4.997304345509695e-05,
      "loss": 1.0251,
      "step": 9389
    },
    {
      "epoch": 0.6768543213436171,
      "grad_norm": 0.15076487530981403,
      "learning_rate": 4.9952829227679597e-05,
      "loss": 1.0299,
      "step": 9390
    },
    {
      "epoch": 0.676926403805954,
      "grad_norm": 0.14004675818572893,
      "learning_rate": 4.9932617728280795e-05,
      "loss": 0.9945,
      "step": 9391
    },
    {
      "epoch": 0.676998486268291,
      "grad_norm": 0.11613355656107972,
      "learning_rate": 4.9912408958002234e-05,
      "loss": 1.0591,
      "step": 9392
    },
    {
      "epoch": 0.6770705687306279,
      "grad_norm": 0.10553528443539138,
      "learning_rate": 4.9892202917945484e-05,
      "loss": 1.0757,
      "step": 9393
    },
    {
      "epoch": 0.6771426511929648,
      "grad_norm": 0.12728955097217937,
      "learning_rate": 4.9871999609211964e-05,
      "loss": 1.0143,
      "step": 9394
    },
    {
      "epoch": 0.6772147336553017,
      "grad_norm": 0.12007918199133472,
      "learning_rate": 4.9851799032902913e-05,
      "loss": 1.0097,
      "step": 9395
    },
    {
      "epoch": 0.6772868161176385,
      "grad_norm": 0.351938474011831,
      "learning_rate": 4.983160119011945e-05,
      "loss": 1.0562,
      "step": 9396
    },
    {
      "epoch": 0.6773588985799754,
      "grad_norm": 0.18106642881333831,
      "learning_rate": 4.9811406081962586e-05,
      "loss": 0.9949,
      "step": 9397
    },
    {
      "epoch": 0.6774309810423124,
      "grad_norm": 0.12862734121945407,
      "learning_rate": 4.9791213709533006e-05,
      "loss": 1.0155,
      "step": 9398
    },
    {
      "epoch": 0.6775030635046493,
      "grad_norm": 0.24747259982494585,
      "learning_rate": 4.97710240739315e-05,
      "loss": 1.0337,
      "step": 9399
    },
    {
      "epoch": 0.6775751459669862,
      "grad_norm": 0.19165408691496602,
      "learning_rate": 4.975083717625855e-05,
      "loss": 1.0001,
      "step": 9400
    },
    {
      "epoch": 0.6776472284293231,
      "grad_norm": 0.13167825581623754,
      "learning_rate": 4.973065301761455e-05,
      "loss": 1.0665,
      "step": 9401
    },
    {
      "epoch": 0.67771931089166,
      "grad_norm": 0.11452152116442722,
      "learning_rate": 4.9710471599099605e-05,
      "loss": 0.988,
      "step": 9402
    },
    {
      "epoch": 0.677791393353997,
      "grad_norm": 0.13857126436392017,
      "learning_rate": 4.969029292181391e-05,
      "loss": 1.0857,
      "step": 9403
    },
    {
      "epoch": 0.6778634758163339,
      "grad_norm": 0.11721310178935027,
      "learning_rate": 4.967011698685732e-05,
      "loss": 1.001,
      "step": 9404
    },
    {
      "epoch": 0.6779355582786708,
      "grad_norm": 0.1206741635618361,
      "learning_rate": 4.9649943795329654e-05,
      "loss": 1.0996,
      "step": 9405
    },
    {
      "epoch": 0.6780076407410077,
      "grad_norm": 0.12411748977007853,
      "learning_rate": 4.962977334833048e-05,
      "loss": 1.0138,
      "step": 9406
    },
    {
      "epoch": 0.6780797232033446,
      "grad_norm": 0.13711934418637872,
      "learning_rate": 4.9609605646959226e-05,
      "loss": 1.084,
      "step": 9407
    },
    {
      "epoch": 0.6781518056656816,
      "grad_norm": 0.14282476091379132,
      "learning_rate": 4.958944069231531e-05,
      "loss": 0.9992,
      "step": 9408
    },
    {
      "epoch": 0.6782238881280185,
      "grad_norm": 0.1174627149487008,
      "learning_rate": 4.9569278485497906e-05,
      "loss": 1.0656,
      "step": 9409
    },
    {
      "epoch": 0.6782959705903554,
      "grad_norm": 0.11949231139076619,
      "learning_rate": 4.954911902760594e-05,
      "loss": 1.1145,
      "step": 9410
    },
    {
      "epoch": 0.6783680530526923,
      "grad_norm": 0.12185611153080478,
      "learning_rate": 4.9528962319738304e-05,
      "loss": 0.9919,
      "step": 9411
    },
    {
      "epoch": 0.6784401355150292,
      "grad_norm": 0.1284504067169777,
      "learning_rate": 4.9508808362993785e-05,
      "loss": 1.1014,
      "step": 9412
    },
    {
      "epoch": 0.6785122179773662,
      "grad_norm": 0.22170917270503,
      "learning_rate": 4.948865715847096e-05,
      "loss": 1.0416,
      "step": 9413
    },
    {
      "epoch": 0.678584300439703,
      "grad_norm": 0.12486860441556012,
      "learning_rate": 4.946850870726817e-05,
      "loss": 1.1185,
      "step": 9414
    },
    {
      "epoch": 0.6786563829020399,
      "grad_norm": 0.21919861217480016,
      "learning_rate": 4.944836301048368e-05,
      "loss": 1.0497,
      "step": 9415
    },
    {
      "epoch": 0.6787284653643768,
      "grad_norm": 0.10513814303083184,
      "learning_rate": 4.9428220069215694e-05,
      "loss": 1.0262,
      "step": 9416
    },
    {
      "epoch": 0.6788005478267137,
      "grad_norm": 0.1565525385055968,
      "learning_rate": 4.940807988456217e-05,
      "loss": 1.0709,
      "step": 9417
    },
    {
      "epoch": 0.6788726302890506,
      "grad_norm": 0.10550534806390673,
      "learning_rate": 4.938794245762088e-05,
      "loss": 1.0005,
      "step": 9418
    },
    {
      "epoch": 0.6789447127513876,
      "grad_norm": 0.11432035550963497,
      "learning_rate": 4.9367807789489504e-05,
      "loss": 1.042,
      "step": 9419
    },
    {
      "epoch": 0.6790167952137245,
      "grad_norm": 0.110557503235671,
      "learning_rate": 4.934767588126554e-05,
      "loss": 1.0206,
      "step": 9420
    },
    {
      "epoch": 0.6790888776760614,
      "grad_norm": 0.19348144012775095,
      "learning_rate": 4.932754673404646e-05,
      "loss": 0.992,
      "step": 9421
    },
    {
      "epoch": 0.6791609601383983,
      "grad_norm": 0.12431032656662817,
      "learning_rate": 4.930742034892938e-05,
      "loss": 1.0535,
      "step": 9422
    },
    {
      "epoch": 0.6792330426007352,
      "grad_norm": 0.10590746855869036,
      "learning_rate": 4.928729672701139e-05,
      "loss": 1.0178,
      "step": 9423
    },
    {
      "epoch": 0.6793051250630722,
      "grad_norm": 0.13737517850258235,
      "learning_rate": 4.926717586938938e-05,
      "loss": 1.0997,
      "step": 9424
    },
    {
      "epoch": 0.6793772075254091,
      "grad_norm": 0.12433161945838095,
      "learning_rate": 4.9247057777160245e-05,
      "loss": 1.0281,
      "step": 9425
    },
    {
      "epoch": 0.679449289987746,
      "grad_norm": 0.1446247061067211,
      "learning_rate": 4.922694245142045e-05,
      "loss": 1.068,
      "step": 9426
    },
    {
      "epoch": 0.6795213724500829,
      "grad_norm": 0.12273150278516774,
      "learning_rate": 4.920682989326654e-05,
      "loss": 1.0298,
      "step": 9427
    },
    {
      "epoch": 0.6795934549124198,
      "grad_norm": 0.11935098403787545,
      "learning_rate": 4.91867201037948e-05,
      "loss": 1.0437,
      "step": 9428
    },
    {
      "epoch": 0.6796655373747568,
      "grad_norm": 0.13351373782732265,
      "learning_rate": 4.91666130841014e-05,
      "loss": 1.0308,
      "step": 9429
    },
    {
      "epoch": 0.6797376198370937,
      "grad_norm": 0.15052092403515177,
      "learning_rate": 4.9146508835282354e-05,
      "loss": 1.1156,
      "step": 9430
    },
    {
      "epoch": 0.6798097022994305,
      "grad_norm": 0.11492260012291947,
      "learning_rate": 4.9126407358433535e-05,
      "loss": 1.1187,
      "step": 9431
    },
    {
      "epoch": 0.6798817847617674,
      "grad_norm": 0.12515539520264535,
      "learning_rate": 4.910630865465063e-05,
      "loss": 0.9783,
      "step": 9432
    },
    {
      "epoch": 0.6799538672241043,
      "grad_norm": 0.10529886928226886,
      "learning_rate": 4.9086212725029226e-05,
      "loss": 0.9724,
      "step": 9433
    },
    {
      "epoch": 0.6800259496864413,
      "grad_norm": 0.15153506604949246,
      "learning_rate": 4.906611957066472e-05,
      "loss": 1.023,
      "step": 9434
    },
    {
      "epoch": 0.6800980321487782,
      "grad_norm": 0.13739369726415052,
      "learning_rate": 4.904602919265234e-05,
      "loss": 1.1479,
      "step": 9435
    },
    {
      "epoch": 0.6801701146111151,
      "grad_norm": 0.11320639859220119,
      "learning_rate": 4.902594159208723e-05,
      "loss": 1.0033,
      "step": 9436
    },
    {
      "epoch": 0.680242197073452,
      "grad_norm": 0.1253745428946434,
      "learning_rate": 4.900585677006434e-05,
      "loss": 1.0434,
      "step": 9437
    },
    {
      "epoch": 0.6803142795357889,
      "grad_norm": 0.17483340192344135,
      "learning_rate": 4.8985774727678446e-05,
      "loss": 1.0638,
      "step": 9438
    },
    {
      "epoch": 0.6803863619981259,
      "grad_norm": 0.16401544545821542,
      "learning_rate": 4.896569546602423e-05,
      "loss": 1.0908,
      "step": 9439
    },
    {
      "epoch": 0.6804584444604628,
      "grad_norm": 0.1182480171401971,
      "learning_rate": 4.8945618986196175e-05,
      "loss": 1.0718,
      "step": 9440
    },
    {
      "epoch": 0.6805305269227997,
      "grad_norm": 0.10898222166729273,
      "learning_rate": 4.892554528928862e-05,
      "loss": 0.9121,
      "step": 9441
    },
    {
      "epoch": 0.6806026093851366,
      "grad_norm": 0.12827674069406517,
      "learning_rate": 4.890547437639578e-05,
      "loss": 1.0374,
      "step": 9442
    },
    {
      "epoch": 0.6806746918474735,
      "grad_norm": 0.1299736307108592,
      "learning_rate": 4.8885406248611686e-05,
      "loss": 1.0454,
      "step": 9443
    },
    {
      "epoch": 0.6807467743098105,
      "grad_norm": 0.10881199535913587,
      "learning_rate": 4.8865340907030246e-05,
      "loss": 1.04,
      "step": 9444
    },
    {
      "epoch": 0.6808188567721474,
      "grad_norm": 0.13499733623646915,
      "learning_rate": 4.8845278352745175e-05,
      "loss": 1.0771,
      "step": 9445
    },
    {
      "epoch": 0.6808909392344843,
      "grad_norm": 0.12291433625237658,
      "learning_rate": 4.8825218586850085e-05,
      "loss": 1.0511,
      "step": 9446
    },
    {
      "epoch": 0.6809630216968212,
      "grad_norm": 0.12282670251110878,
      "learning_rate": 4.880516161043841e-05,
      "loss": 1.0652,
      "step": 9447
    },
    {
      "epoch": 0.681035104159158,
      "grad_norm": 0.10706833399684691,
      "learning_rate": 4.8785107424603414e-05,
      "loss": 1.0941,
      "step": 9448
    },
    {
      "epoch": 0.6811071866214949,
      "grad_norm": 0.12993669782520767,
      "learning_rate": 4.876505603043826e-05,
      "loss": 1.0342,
      "step": 9449
    },
    {
      "epoch": 0.6811792690838319,
      "grad_norm": 0.13231855172970788,
      "learning_rate": 4.874500742903595e-05,
      "loss": 0.9987,
      "step": 9450
    },
    {
      "epoch": 0.6812513515461688,
      "grad_norm": 0.10672064896880698,
      "learning_rate": 4.8724961621489194e-05,
      "loss": 1.0893,
      "step": 9451
    },
    {
      "epoch": 0.6813234340085057,
      "grad_norm": 0.29074341910569085,
      "learning_rate": 4.870491860889079e-05,
      "loss": 1.0372,
      "step": 9452
    },
    {
      "epoch": 0.6813955164708426,
      "grad_norm": 0.16777724771040833,
      "learning_rate": 4.8684878392333225e-05,
      "loss": 0.9916,
      "step": 9453
    },
    {
      "epoch": 0.6814675989331795,
      "grad_norm": 0.12790102263633657,
      "learning_rate": 4.8664840972908885e-05,
      "loss": 1.0719,
      "step": 9454
    },
    {
      "epoch": 0.6815396813955165,
      "grad_norm": 0.10810775336035496,
      "learning_rate": 4.8644806351709915e-05,
      "loss": 0.9979,
      "step": 9455
    },
    {
      "epoch": 0.6816117638578534,
      "grad_norm": 0.11926632231467786,
      "learning_rate": 4.8624774529828464e-05,
      "loss": 0.9216,
      "step": 9456
    },
    {
      "epoch": 0.6816838463201903,
      "grad_norm": 0.12511821519102895,
      "learning_rate": 4.860474550835642e-05,
      "loss": 1.0001,
      "step": 9457
    },
    {
      "epoch": 0.6817559287825272,
      "grad_norm": 0.11702487208170624,
      "learning_rate": 4.8584719288385575e-05,
      "loss": 1.0906,
      "step": 9458
    },
    {
      "epoch": 0.6818280112448641,
      "grad_norm": 0.11874007268035833,
      "learning_rate": 4.856469587100743e-05,
      "loss": 1.0118,
      "step": 9459
    },
    {
      "epoch": 0.6819000937072011,
      "grad_norm": 0.12241534296231701,
      "learning_rate": 4.8544675257313554e-05,
      "loss": 1.0667,
      "step": 9460
    },
    {
      "epoch": 0.681972176169538,
      "grad_norm": 0.11196659926658804,
      "learning_rate": 4.85246574483952e-05,
      "loss": 1.0612,
      "step": 9461
    },
    {
      "epoch": 0.6820442586318749,
      "grad_norm": 0.15310087132637126,
      "learning_rate": 4.8504642445343575e-05,
      "loss": 1.1008,
      "step": 9462
    },
    {
      "epoch": 0.6821163410942118,
      "grad_norm": 0.14107032449983528,
      "learning_rate": 4.8484630249249584e-05,
      "loss": 0.9794,
      "step": 9463
    },
    {
      "epoch": 0.6821884235565487,
      "grad_norm": 0.11520289444319812,
      "learning_rate": 4.846462086120407e-05,
      "loss": 1.0521,
      "step": 9464
    },
    {
      "epoch": 0.6822605060188857,
      "grad_norm": 0.11874568436639696,
      "learning_rate": 4.844461428229782e-05,
      "loss": 1.0449,
      "step": 9465
    },
    {
      "epoch": 0.6823325884812225,
      "grad_norm": 0.11275724672711705,
      "learning_rate": 4.842461051362135e-05,
      "loss": 1.0424,
      "step": 9466
    },
    {
      "epoch": 0.6824046709435594,
      "grad_norm": 0.1211666328815643,
      "learning_rate": 4.8404609556264967e-05,
      "loss": 1.0152,
      "step": 9467
    },
    {
      "epoch": 0.6824767534058963,
      "grad_norm": 0.10623852569895688,
      "learning_rate": 4.838461141131892e-05,
      "loss": 1.01,
      "step": 9468
    },
    {
      "epoch": 0.6825488358682332,
      "grad_norm": 0.12174438312911748,
      "learning_rate": 4.836461607987336e-05,
      "loss": 1.0559,
      "step": 9469
    },
    {
      "epoch": 0.6826209183305701,
      "grad_norm": 0.14449841657203868,
      "learning_rate": 4.8344623563018186e-05,
      "loss": 1.0489,
      "step": 9470
    },
    {
      "epoch": 0.6826930007929071,
      "grad_norm": 0.1245691363165646,
      "learning_rate": 4.832463386184313e-05,
      "loss": 1.0147,
      "step": 9471
    },
    {
      "epoch": 0.682765083255244,
      "grad_norm": 0.13486143286757582,
      "learning_rate": 4.830464697743783e-05,
      "loss": 1.0415,
      "step": 9472
    },
    {
      "epoch": 0.6828371657175809,
      "grad_norm": 0.12304072390579386,
      "learning_rate": 4.82846629108917e-05,
      "loss": 0.962,
      "step": 9473
    },
    {
      "epoch": 0.6829092481799178,
      "grad_norm": 0.10687504854061004,
      "learning_rate": 4.8264681663294174e-05,
      "loss": 1.045,
      "step": 9474
    },
    {
      "epoch": 0.6829813306422547,
      "grad_norm": 0.12434775486001325,
      "learning_rate": 4.8244703235734315e-05,
      "loss": 1.053,
      "step": 9475
    },
    {
      "epoch": 0.6830534131045917,
      "grad_norm": 0.13436693153655777,
      "learning_rate": 4.822472762930112e-05,
      "loss": 1.0578,
      "step": 9476
    },
    {
      "epoch": 0.6831254955669286,
      "grad_norm": 0.991427981018273,
      "learning_rate": 4.820475484508345e-05,
      "loss": 1.0031,
      "step": 9477
    },
    {
      "epoch": 0.6831975780292655,
      "grad_norm": 0.1329527502312143,
      "learning_rate": 4.818478488417007e-05,
      "loss": 1.0528,
      "step": 9478
    },
    {
      "epoch": 0.6832696604916024,
      "grad_norm": 0.12122706325194327,
      "learning_rate": 4.816481774764945e-05,
      "loss": 1.057,
      "step": 9479
    },
    {
      "epoch": 0.6833417429539393,
      "grad_norm": 0.15349867769572842,
      "learning_rate": 4.814485343660998e-05,
      "loss": 1.0481,
      "step": 9480
    },
    {
      "epoch": 0.6834138254162763,
      "grad_norm": 0.15601632964007175,
      "learning_rate": 4.812489195213987e-05,
      "loss": 1.021,
      "step": 9481
    },
    {
      "epoch": 0.6834859078786132,
      "grad_norm": 0.17887203797679832,
      "learning_rate": 4.810493329532731e-05,
      "loss": 1.0544,
      "step": 9482
    },
    {
      "epoch": 0.68355799034095,
      "grad_norm": 0.11492740863431823,
      "learning_rate": 4.808497746726013e-05,
      "loss": 1.0888,
      "step": 9483
    },
    {
      "epoch": 0.6836300728032869,
      "grad_norm": 0.8360504990748265,
      "learning_rate": 4.80650244690261e-05,
      "loss": 1.0908,
      "step": 9484
    },
    {
      "epoch": 0.6837021552656238,
      "grad_norm": 0.11789866593261215,
      "learning_rate": 4.804507430171288e-05,
      "loss": 1.0711,
      "step": 9485
    },
    {
      "epoch": 0.6837742377279608,
      "grad_norm": 0.3796689378289281,
      "learning_rate": 4.8025126966407896e-05,
      "loss": 0.9964,
      "step": 9486
    },
    {
      "epoch": 0.6838463201902977,
      "grad_norm": 0.11557646844707575,
      "learning_rate": 4.800518246419849e-05,
      "loss": 1.0436,
      "step": 9487
    },
    {
      "epoch": 0.6839184026526346,
      "grad_norm": 0.12017240154141488,
      "learning_rate": 4.798524079617178e-05,
      "loss": 1.0596,
      "step": 9488
    },
    {
      "epoch": 0.6839904851149715,
      "grad_norm": 0.13610859349687507,
      "learning_rate": 4.79653019634148e-05,
      "loss": 1.0792,
      "step": 9489
    },
    {
      "epoch": 0.6840625675773084,
      "grad_norm": 0.17855499267298278,
      "learning_rate": 4.7945365967014366e-05,
      "loss": 0.9998,
      "step": 9490
    },
    {
      "epoch": 0.6841346500396454,
      "grad_norm": 0.11829564998752029,
      "learning_rate": 4.792543280805719e-05,
      "loss": 1.0085,
      "step": 9491
    },
    {
      "epoch": 0.6842067325019823,
      "grad_norm": 0.13040077568835462,
      "learning_rate": 4.7905502487629796e-05,
      "loss": 1.1815,
      "step": 9492
    },
    {
      "epoch": 0.6842788149643192,
      "grad_norm": 0.12593552335586108,
      "learning_rate": 4.788557500681856e-05,
      "loss": 1.1201,
      "step": 9493
    },
    {
      "epoch": 0.6843508974266561,
      "grad_norm": 0.1269421097397982,
      "learning_rate": 4.786565036670972e-05,
      "loss": 1.0703,
      "step": 9494
    },
    {
      "epoch": 0.684422979888993,
      "grad_norm": 0.11891005761452944,
      "learning_rate": 4.784572856838935e-05,
      "loss": 1.0575,
      "step": 9495
    },
    {
      "epoch": 0.68449506235133,
      "grad_norm": 0.1301967693987817,
      "learning_rate": 4.782580961294335e-05,
      "loss": 1.0931,
      "step": 9496
    },
    {
      "epoch": 0.6845671448136669,
      "grad_norm": 0.10446205870061145,
      "learning_rate": 4.780589350145749e-05,
      "loss": 1.0323,
      "step": 9497
    },
    {
      "epoch": 0.6846392272760038,
      "grad_norm": 0.1324010872400509,
      "learning_rate": 4.778598023501738e-05,
      "loss": 1.0226,
      "step": 9498
    },
    {
      "epoch": 0.6847113097383407,
      "grad_norm": 0.2236271844318049,
      "learning_rate": 4.776606981470846e-05,
      "loss": 1.0537,
      "step": 9499
    },
    {
      "epoch": 0.6847833922006775,
      "grad_norm": 0.12042859883034972,
      "learning_rate": 4.774616224161604e-05,
      "loss": 1.0148,
      "step": 9500
    },
    {
      "epoch": 0.6848554746630144,
      "grad_norm": 0.12986139497778368,
      "learning_rate": 4.7726257516825243e-05,
      "loss": 1.0906,
      "step": 9501
    },
    {
      "epoch": 0.6849275571253514,
      "grad_norm": 2.062737479072257,
      "learning_rate": 4.7706355641421084e-05,
      "loss": 1.097,
      "step": 9502
    },
    {
      "epoch": 0.6849996395876883,
      "grad_norm": 0.12640486469386966,
      "learning_rate": 4.7686456616488365e-05,
      "loss": 1.0727,
      "step": 9503
    },
    {
      "epoch": 0.6850717220500252,
      "grad_norm": 0.12525283418142105,
      "learning_rate": 4.7666560443111774e-05,
      "loss": 1.0128,
      "step": 9504
    },
    {
      "epoch": 0.6851438045123621,
      "grad_norm": 0.5037448814408423,
      "learning_rate": 4.764666712237582e-05,
      "loss": 0.9959,
      "step": 9505
    },
    {
      "epoch": 0.685215886974699,
      "grad_norm": 0.13559311804918003,
      "learning_rate": 4.762677665536488e-05,
      "loss": 1.0912,
      "step": 9506
    },
    {
      "epoch": 0.685287969437036,
      "grad_norm": 0.15632804644031906,
      "learning_rate": 4.7606889043163196e-05,
      "loss": 1.0571,
      "step": 9507
    },
    {
      "epoch": 0.6853600518993729,
      "grad_norm": 0.12494510725059366,
      "learning_rate": 4.7587004286854695e-05,
      "loss": 1.0259,
      "step": 9508
    },
    {
      "epoch": 0.6854321343617098,
      "grad_norm": 0.20456869573545827,
      "learning_rate": 4.756712238752341e-05,
      "loss": 1.081,
      "step": 9509
    },
    {
      "epoch": 0.6855042168240467,
      "grad_norm": 0.12508099965911396,
      "learning_rate": 4.7547243346253024e-05,
      "loss": 1.0194,
      "step": 9510
    },
    {
      "epoch": 0.6855762992863836,
      "grad_norm": 0.11545241449872591,
      "learning_rate": 4.752736716412718e-05,
      "loss": 1.0829,
      "step": 9511
    },
    {
      "epoch": 0.6856483817487206,
      "grad_norm": 0.1158319398088618,
      "learning_rate": 4.750749384222918e-05,
      "loss": 1.0576,
      "step": 9512
    },
    {
      "epoch": 0.6857204642110575,
      "grad_norm": 0.12826829375925194,
      "learning_rate": 4.748762338164241e-05,
      "loss": 0.988,
      "step": 9513
    },
    {
      "epoch": 0.6857925466733944,
      "grad_norm": 0.12102649664314191,
      "learning_rate": 4.7467755783449955e-05,
      "loss": 1.0286,
      "step": 9514
    },
    {
      "epoch": 0.6858646291357313,
      "grad_norm": 0.12155618624238217,
      "learning_rate": 4.744789104873482e-05,
      "loss": 0.9762,
      "step": 9515
    },
    {
      "epoch": 0.6859367115980682,
      "grad_norm": 0.1255375724263086,
      "learning_rate": 4.7428029178579726e-05,
      "loss": 0.9832,
      "step": 9516
    },
    {
      "epoch": 0.6860087940604052,
      "grad_norm": 0.13070943701927915,
      "learning_rate": 4.7408170174067326e-05,
      "loss": 1.084,
      "step": 9517
    },
    {
      "epoch": 0.686080876522742,
      "grad_norm": 0.16183731151742248,
      "learning_rate": 4.7388314036280224e-05,
      "loss": 1.0914,
      "step": 9518
    },
    {
      "epoch": 0.6861529589850789,
      "grad_norm": 0.40276757176672734,
      "learning_rate": 4.736846076630065e-05,
      "loss": 1.0163,
      "step": 9519
    },
    {
      "epoch": 0.6862250414474158,
      "grad_norm": 0.1322855020375832,
      "learning_rate": 4.734861036521082e-05,
      "loss": 0.9926,
      "step": 9520
    },
    {
      "epoch": 0.6862971239097527,
      "grad_norm": 0.11389746088790327,
      "learning_rate": 4.732876283409272e-05,
      "loss": 1.0108,
      "step": 9521
    },
    {
      "epoch": 0.6863692063720896,
      "grad_norm": 0.11955612547717707,
      "learning_rate": 4.7308918174028326e-05,
      "loss": 0.984,
      "step": 9522
    },
    {
      "epoch": 0.6864412888344266,
      "grad_norm": 0.11114435351503227,
      "learning_rate": 4.728907638609925e-05,
      "loss": 0.9538,
      "step": 9523
    },
    {
      "epoch": 0.6865133712967635,
      "grad_norm": 0.12088637056939129,
      "learning_rate": 4.7269237471387075e-05,
      "loss": 1.0478,
      "step": 9524
    },
    {
      "epoch": 0.6865854537591004,
      "grad_norm": 0.11128760342702444,
      "learning_rate": 4.724940143097316e-05,
      "loss": 0.9988,
      "step": 9525
    },
    {
      "epoch": 0.6866575362214373,
      "grad_norm": 0.1252089014696476,
      "learning_rate": 4.722956826593886e-05,
      "loss": 1.0944,
      "step": 9526
    },
    {
      "epoch": 0.6867296186837742,
      "grad_norm": 0.11788142942096964,
      "learning_rate": 4.720973797736515e-05,
      "loss": 1.0787,
      "step": 9527
    },
    {
      "epoch": 0.6868017011461112,
      "grad_norm": 0.11115066762428727,
      "learning_rate": 4.7189910566333006e-05,
      "loss": 1.0044,
      "step": 9528
    },
    {
      "epoch": 0.6868737836084481,
      "grad_norm": 0.1342649076721258,
      "learning_rate": 4.717008603392318e-05,
      "loss": 1.0098,
      "step": 9529
    },
    {
      "epoch": 0.686945866070785,
      "grad_norm": 0.11330291484214548,
      "learning_rate": 4.715026438121629e-05,
      "loss": 0.9908,
      "step": 9530
    },
    {
      "epoch": 0.6870179485331219,
      "grad_norm": 0.13352593420115225,
      "learning_rate": 4.71304456092928e-05,
      "loss": 1.0509,
      "step": 9531
    },
    {
      "epoch": 0.6870900309954588,
      "grad_norm": 0.11133546168916109,
      "learning_rate": 4.7110629719233025e-05,
      "loss": 1.0468,
      "step": 9532
    },
    {
      "epoch": 0.6871621134577958,
      "grad_norm": 0.20225902766149767,
      "learning_rate": 4.709081671211708e-05,
      "loss": 1.0898,
      "step": 9533
    },
    {
      "epoch": 0.6872341959201327,
      "grad_norm": 0.11777734858844541,
      "learning_rate": 4.707100658902497e-05,
      "loss": 1.0829,
      "step": 9534
    },
    {
      "epoch": 0.6873062783824695,
      "grad_norm": 0.11789795156189817,
      "learning_rate": 4.70511993510365e-05,
      "loss": 0.8863,
      "step": 9535
    },
    {
      "epoch": 0.6873783608448064,
      "grad_norm": 0.11257883014268918,
      "learning_rate": 4.703139499923138e-05,
      "loss": 0.95,
      "step": 9536
    },
    {
      "epoch": 0.6874504433071433,
      "grad_norm": 0.18311878639696405,
      "learning_rate": 4.701159353468909e-05,
      "loss": 1.1166,
      "step": 9537
    },
    {
      "epoch": 0.6875225257694803,
      "grad_norm": 0.12370744935582316,
      "learning_rate": 4.6991794958489e-05,
      "loss": 1.1038,
      "step": 9538
    },
    {
      "epoch": 0.6875946082318172,
      "grad_norm": 0.11178537554781529,
      "learning_rate": 4.697199927171032e-05,
      "loss": 1.0854,
      "step": 9539
    },
    {
      "epoch": 0.6876666906941541,
      "grad_norm": 0.12124353108403065,
      "learning_rate": 4.695220647543207e-05,
      "loss": 0.9873,
      "step": 9540
    },
    {
      "epoch": 0.687738773156491,
      "grad_norm": 0.10649705983554995,
      "learning_rate": 4.693241657073315e-05,
      "loss": 1.1321,
      "step": 9541
    },
    {
      "epoch": 0.6878108556188279,
      "grad_norm": 0.10655431217301263,
      "learning_rate": 4.691262955869229e-05,
      "loss": 0.9759,
      "step": 9542
    },
    {
      "epoch": 0.6878829380811649,
      "grad_norm": 0.1168025306495649,
      "learning_rate": 4.6892845440388044e-05,
      "loss": 1.0296,
      "step": 9543
    },
    {
      "epoch": 0.6879550205435018,
      "grad_norm": 0.11927813433835849,
      "learning_rate": 4.687306421689883e-05,
      "loss": 0.9652,
      "step": 9544
    },
    {
      "epoch": 0.6880271030058387,
      "grad_norm": 0.14823128675848807,
      "learning_rate": 4.685328588930291e-05,
      "loss": 1.0198,
      "step": 9545
    },
    {
      "epoch": 0.6880991854681756,
      "grad_norm": 0.10712999692935415,
      "learning_rate": 4.6833510458678375e-05,
      "loss": 1.1065,
      "step": 9546
    },
    {
      "epoch": 0.6881712679305125,
      "grad_norm": 0.1197195413215674,
      "learning_rate": 4.681373792610315e-05,
      "loss": 1.0276,
      "step": 9547
    },
    {
      "epoch": 0.6882433503928495,
      "grad_norm": 0.110622482279534,
      "learning_rate": 4.6793968292655045e-05,
      "loss": 0.9817,
      "step": 9548
    },
    {
      "epoch": 0.6883154328551864,
      "grad_norm": 0.10924904320831015,
      "learning_rate": 4.677420155941165e-05,
      "loss": 0.9939,
      "step": 9549
    },
    {
      "epoch": 0.6883875153175233,
      "grad_norm": 0.16118333999647205,
      "learning_rate": 4.675443772745044e-05,
      "loss": 1.0116,
      "step": 9550
    },
    {
      "epoch": 0.6884595977798602,
      "grad_norm": 0.12167607831472269,
      "learning_rate": 4.673467679784875e-05,
      "loss": 1.0275,
      "step": 9551
    },
    {
      "epoch": 0.688531680242197,
      "grad_norm": 0.12146308461620575,
      "learning_rate": 4.6714918771683646e-05,
      "loss": 1.0286,
      "step": 9552
    },
    {
      "epoch": 0.6886037627045339,
      "grad_norm": 0.11756987090662242,
      "learning_rate": 4.6695163650032205e-05,
      "loss": 1.0576,
      "step": 9553
    },
    {
      "epoch": 0.6886758451668709,
      "grad_norm": 0.11042286638743615,
      "learning_rate": 4.6675411433971205e-05,
      "loss": 1.0284,
      "step": 9554
    },
    {
      "epoch": 0.6887479276292078,
      "grad_norm": 0.1222843768814662,
      "learning_rate": 4.6655662124577395e-05,
      "loss": 1.0344,
      "step": 9555
    },
    {
      "epoch": 0.6888200100915447,
      "grad_norm": 0.2450087177126903,
      "learning_rate": 4.663591572292715e-05,
      "loss": 1.0197,
      "step": 9556
    },
    {
      "epoch": 0.6888920925538816,
      "grad_norm": 0.14052475399677278,
      "learning_rate": 4.661617223009695e-05,
      "loss": 1.1175,
      "step": 9557
    },
    {
      "epoch": 0.6889641750162185,
      "grad_norm": 0.14148825434114864,
      "learning_rate": 4.659643164716293e-05,
      "loss": 1.1321,
      "step": 9558
    },
    {
      "epoch": 0.6890362574785555,
      "grad_norm": 0.15037861147282772,
      "learning_rate": 4.65766939752012e-05,
      "loss": 1.004,
      "step": 9559
    },
    {
      "epoch": 0.6891083399408924,
      "grad_norm": 0.11970576436646589,
      "learning_rate": 4.655695921528754e-05,
      "loss": 1.0834,
      "step": 9560
    },
    {
      "epoch": 0.6891804224032293,
      "grad_norm": 0.12481347679249803,
      "learning_rate": 4.653722736849769e-05,
      "loss": 1.0665,
      "step": 9561
    },
    {
      "epoch": 0.6892525048655662,
      "grad_norm": 0.12278233967462328,
      "learning_rate": 4.651749843590728e-05,
      "loss": 1.0961,
      "step": 9562
    },
    {
      "epoch": 0.6893245873279031,
      "grad_norm": 0.11240364632726936,
      "learning_rate": 4.64977724185917e-05,
      "loss": 1.0525,
      "step": 9563
    },
    {
      "epoch": 0.6893966697902401,
      "grad_norm": 0.11086146527396033,
      "learning_rate": 4.647804931762614e-05,
      "loss": 1.0713,
      "step": 9564
    },
    {
      "epoch": 0.689468752252577,
      "grad_norm": 0.12272550415871109,
      "learning_rate": 4.645832913408566e-05,
      "loss": 1.0769,
      "step": 9565
    },
    {
      "epoch": 0.6895408347149139,
      "grad_norm": 0.11777977890760294,
      "learning_rate": 4.6438611869045304e-05,
      "loss": 1.0484,
      "step": 9566
    },
    {
      "epoch": 0.6896129171772508,
      "grad_norm": 0.1206716139011292,
      "learning_rate": 4.6418897523579785e-05,
      "loss": 1.124,
      "step": 9567
    },
    {
      "epoch": 0.6896849996395877,
      "grad_norm": 0.15304053829377431,
      "learning_rate": 4.639918609876368e-05,
      "loss": 1.0742,
      "step": 9568
    },
    {
      "epoch": 0.6897570821019247,
      "grad_norm": 0.11805383719443227,
      "learning_rate": 4.6379477595671416e-05,
      "loss": 1.0374,
      "step": 9569
    },
    {
      "epoch": 0.6898291645642615,
      "grad_norm": 0.1315430920229178,
      "learning_rate": 4.635977201537737e-05,
      "loss": 0.9966,
      "step": 9570
    },
    {
      "epoch": 0.6899012470265984,
      "grad_norm": 0.11000803634129715,
      "learning_rate": 4.634006935895565e-05,
      "loss": 1.1225,
      "step": 9571
    },
    {
      "epoch": 0.6899733294889353,
      "grad_norm": 0.11494981412034651,
      "learning_rate": 4.6320369627480184e-05,
      "loss": 1.0079,
      "step": 9572
    },
    {
      "epoch": 0.6900454119512722,
      "grad_norm": 0.10358349704009356,
      "learning_rate": 4.630067282202479e-05,
      "loss": 0.9899,
      "step": 9573
    },
    {
      "epoch": 0.6901174944136091,
      "grad_norm": 0.1207204804560799,
      "learning_rate": 4.628097894366311e-05,
      "loss": 1.0385,
      "step": 9574
    },
    {
      "epoch": 0.6901895768759461,
      "grad_norm": 0.12994158403819078,
      "learning_rate": 4.6261287993468737e-05,
      "loss": 1.0371,
      "step": 9575
    },
    {
      "epoch": 0.690261659338283,
      "grad_norm": 0.1220218452015431,
      "learning_rate": 4.6241599972514893e-05,
      "loss": 1.0897,
      "step": 9576
    },
    {
      "epoch": 0.6903337418006199,
      "grad_norm": 0.11114689692676281,
      "learning_rate": 4.6221914881874784e-05,
      "loss": 1.0625,
      "step": 9577
    },
    {
      "epoch": 0.6904058242629568,
      "grad_norm": 0.11115448447890003,
      "learning_rate": 4.620223272262139e-05,
      "loss": 1.0508,
      "step": 9578
    },
    {
      "epoch": 0.6904779067252937,
      "grad_norm": 0.13425428522514518,
      "learning_rate": 4.618255349582766e-05,
      "loss": 1.0483,
      "step": 9579
    },
    {
      "epoch": 0.6905499891876307,
      "grad_norm": 0.11519687862497273,
      "learning_rate": 4.616287720256621e-05,
      "loss": 1.0498,
      "step": 9580
    },
    {
      "epoch": 0.6906220716499676,
      "grad_norm": 0.1155236667427514,
      "learning_rate": 4.614320384390959e-05,
      "loss": 0.9979,
      "step": 9581
    },
    {
      "epoch": 0.6906941541123045,
      "grad_norm": 0.8075046368991052,
      "learning_rate": 4.612353342093017e-05,
      "loss": 1.0213,
      "step": 9582
    },
    {
      "epoch": 0.6907662365746414,
      "grad_norm": 0.12467697511463839,
      "learning_rate": 4.6103865934700175e-05,
      "loss": 1.0232,
      "step": 9583
    },
    {
      "epoch": 0.6908383190369783,
      "grad_norm": 0.13675270215321417,
      "learning_rate": 4.608420138629166e-05,
      "loss": 1.06,
      "step": 9584
    },
    {
      "epoch": 0.6909104014993153,
      "grad_norm": 0.11289580650087108,
      "learning_rate": 4.60645397767765e-05,
      "loss": 1.0112,
      "step": 9585
    },
    {
      "epoch": 0.6909824839616522,
      "grad_norm": 0.7524334530375704,
      "learning_rate": 4.6044881107226455e-05,
      "loss": 1.1589,
      "step": 9586
    },
    {
      "epoch": 0.691054566423989,
      "grad_norm": 0.11887498375959848,
      "learning_rate": 4.602522537871309e-05,
      "loss": 1.1156,
      "step": 9587
    },
    {
      "epoch": 0.6911266488863259,
      "grad_norm": 0.15130958961676902,
      "learning_rate": 4.60055725923078e-05,
      "loss": 1.0499,
      "step": 9588
    },
    {
      "epoch": 0.6911987313486628,
      "grad_norm": 0.12009150028065037,
      "learning_rate": 4.5985922749081864e-05,
      "loss": 1.066,
      "step": 9589
    },
    {
      "epoch": 0.6912708138109998,
      "grad_norm": 0.12160641854141332,
      "learning_rate": 4.5966275850106354e-05,
      "loss": 1.0277,
      "step": 9590
    },
    {
      "epoch": 0.6913428962733367,
      "grad_norm": 0.15404326827346598,
      "learning_rate": 4.594663189645222e-05,
      "loss": 1.0065,
      "step": 9591
    },
    {
      "epoch": 0.6914149787356736,
      "grad_norm": 0.14041371265893302,
      "learning_rate": 4.592699088919021e-05,
      "loss": 1.0692,
      "step": 9592
    },
    {
      "epoch": 0.6914870611980105,
      "grad_norm": 0.10518366309201362,
      "learning_rate": 4.590735282939096e-05,
      "loss": 1.014,
      "step": 9593
    },
    {
      "epoch": 0.6915591436603474,
      "grad_norm": 0.13832272189207467,
      "learning_rate": 4.588771771812489e-05,
      "loss": 1.0495,
      "step": 9594
    },
    {
      "epoch": 0.6916312261226844,
      "grad_norm": 0.114836391597436,
      "learning_rate": 4.586808555646233e-05,
      "loss": 1.0247,
      "step": 9595
    },
    {
      "epoch": 0.6917033085850213,
      "grad_norm": 0.3045770749649137,
      "learning_rate": 4.5848456345473365e-05,
      "loss": 1.0447,
      "step": 9596
    },
    {
      "epoch": 0.6917753910473582,
      "grad_norm": 0.228640926348656,
      "learning_rate": 4.5828830086227994e-05,
      "loss": 1.0075,
      "step": 9597
    },
    {
      "epoch": 0.6918474735096951,
      "grad_norm": 0.1268027772686017,
      "learning_rate": 4.5809206779796e-05,
      "loss": 1.0084,
      "step": 9598
    },
    {
      "epoch": 0.691919555972032,
      "grad_norm": 0.13030892396599092,
      "learning_rate": 4.578958642724706e-05,
      "loss": 1.0687,
      "step": 9599
    },
    {
      "epoch": 0.691991638434369,
      "grad_norm": 0.1296131262246862,
      "learning_rate": 4.5769969029650625e-05,
      "loss": 1.0337,
      "step": 9600
    },
    {
      "epoch": 0.6920637208967059,
      "grad_norm": 0.1192956325907379,
      "learning_rate": 4.575035458807604e-05,
      "loss": 1.05,
      "step": 9601
    },
    {
      "epoch": 0.6921358033590428,
      "grad_norm": 0.31234954153887007,
      "learning_rate": 4.573074310359246e-05,
      "loss": 1.0072,
      "step": 9602
    },
    {
      "epoch": 0.6922078858213797,
      "grad_norm": 0.13269150661225393,
      "learning_rate": 4.5711134577268874e-05,
      "loss": 1.0305,
      "step": 9603
    },
    {
      "epoch": 0.6922799682837165,
      "grad_norm": 0.15444765577933794,
      "learning_rate": 4.5691529010174194e-05,
      "loss": 1.0517,
      "step": 9604
    },
    {
      "epoch": 0.6923520507460534,
      "grad_norm": 0.207670751323989,
      "learning_rate": 4.5671926403376954e-05,
      "loss": 1.0058,
      "step": 9605
    },
    {
      "epoch": 0.6924241332083904,
      "grad_norm": 0.12913207281273267,
      "learning_rate": 4.5652326757945804e-05,
      "loss": 1.0921,
      "step": 9606
    },
    {
      "epoch": 0.6924962156707273,
      "grad_norm": 0.12319603712682888,
      "learning_rate": 4.563273007494905e-05,
      "loss": 1.0047,
      "step": 9607
    },
    {
      "epoch": 0.6925682981330642,
      "grad_norm": 0.12281117780931405,
      "learning_rate": 4.561313635545492e-05,
      "loss": 1.064,
      "step": 9608
    },
    {
      "epoch": 0.6926403805954011,
      "grad_norm": 0.130544278589486,
      "learning_rate": 4.559354560053135e-05,
      "loss": 1.0668,
      "step": 9609
    },
    {
      "epoch": 0.692712463057738,
      "grad_norm": 0.1431024974006158,
      "learning_rate": 4.557395781124632e-05,
      "loss": 1.0605,
      "step": 9610
    },
    {
      "epoch": 0.692784545520075,
      "grad_norm": 0.1431021610797228,
      "learning_rate": 4.555437298866749e-05,
      "loss": 1.1293,
      "step": 9611
    },
    {
      "epoch": 0.6928566279824119,
      "grad_norm": 0.11871948144233056,
      "learning_rate": 4.553479113386246e-05,
      "loss": 1.0273,
      "step": 9612
    },
    {
      "epoch": 0.6929287104447488,
      "grad_norm": 0.1131751670612407,
      "learning_rate": 4.55152122478985e-05,
      "loss": 1.0131,
      "step": 9613
    },
    {
      "epoch": 0.6930007929070857,
      "grad_norm": 0.19474207713724564,
      "learning_rate": 4.549563633184294e-05,
      "loss": 1.0802,
      "step": 9614
    },
    {
      "epoch": 0.6930728753694226,
      "grad_norm": 0.11684980717554715,
      "learning_rate": 4.54760633867628e-05,
      "loss": 1.0564,
      "step": 9615
    },
    {
      "epoch": 0.6931449578317596,
      "grad_norm": 0.1290956828505784,
      "learning_rate": 4.5456493413725046e-05,
      "loss": 0.9818,
      "step": 9616
    },
    {
      "epoch": 0.6932170402940965,
      "grad_norm": 0.12243411391695425,
      "learning_rate": 4.543692641379631e-05,
      "loss": 1.084,
      "step": 9617
    },
    {
      "epoch": 0.6932891227564334,
      "grad_norm": 0.1388910321798441,
      "learning_rate": 4.5417362388043194e-05,
      "loss": 1.0739,
      "step": 9618
    },
    {
      "epoch": 0.6933612052187703,
      "grad_norm": 0.13056081401846953,
      "learning_rate": 4.5397801337532164e-05,
      "loss": 0.979,
      "step": 9619
    },
    {
      "epoch": 0.6934332876811072,
      "grad_norm": 0.12588925935240208,
      "learning_rate": 4.537824326332949e-05,
      "loss": 1.0653,
      "step": 9620
    },
    {
      "epoch": 0.6935053701434442,
      "grad_norm": 0.13660668952367877,
      "learning_rate": 4.535868816650118e-05,
      "loss": 1.0708,
      "step": 9621
    },
    {
      "epoch": 0.693577452605781,
      "grad_norm": 0.10587898641373843,
      "learning_rate": 4.533913604811317e-05,
      "loss": 1.0542,
      "step": 9622
    },
    {
      "epoch": 0.6936495350681179,
      "grad_norm": 0.12507966344290267,
      "learning_rate": 4.531958690923128e-05,
      "loss": 1.0099,
      "step": 9623
    },
    {
      "epoch": 0.6937216175304548,
      "grad_norm": 0.1213865646505587,
      "learning_rate": 4.5300040750921135e-05,
      "loss": 1.0174,
      "step": 9624
    },
    {
      "epoch": 0.6937936999927917,
      "grad_norm": 0.17986591599413967,
      "learning_rate": 4.5280497574248094e-05,
      "loss": 1.0085,
      "step": 9625
    },
    {
      "epoch": 0.6938657824551286,
      "grad_norm": 0.11406443900373568,
      "learning_rate": 4.5260957380277456e-05,
      "loss": 1.0962,
      "step": 9626
    },
    {
      "epoch": 0.6939378649174656,
      "grad_norm": 0.10269719873797033,
      "learning_rate": 4.524142017007432e-05,
      "loss": 1.0381,
      "step": 9627
    },
    {
      "epoch": 0.6940099473798025,
      "grad_norm": 0.12165273822050114,
      "learning_rate": 4.522188594470375e-05,
      "loss": 0.9675,
      "step": 9628
    },
    {
      "epoch": 0.6940820298421394,
      "grad_norm": 0.12388644955544204,
      "learning_rate": 4.520235470523041e-05,
      "loss": 1.0478,
      "step": 9629
    },
    {
      "epoch": 0.6941541123044763,
      "grad_norm": 0.12315896495600523,
      "learning_rate": 4.518282645271897e-05,
      "loss": 1.0982,
      "step": 9630
    },
    {
      "epoch": 0.6942261947668132,
      "grad_norm": 0.128402172825118,
      "learning_rate": 4.516330118823387e-05,
      "loss": 0.978,
      "step": 9631
    },
    {
      "epoch": 0.6942982772291502,
      "grad_norm": 0.12312947954571087,
      "learning_rate": 4.5143778912839506e-05,
      "loss": 1.0599,
      "step": 9632
    },
    {
      "epoch": 0.6943703596914871,
      "grad_norm": 0.15662500106750502,
      "learning_rate": 4.512425962759992e-05,
      "loss": 1.0568,
      "step": 9633
    },
    {
      "epoch": 0.694442442153824,
      "grad_norm": 0.11548362073952807,
      "learning_rate": 4.5104743333579104e-05,
      "loss": 1.0456,
      "step": 9634
    },
    {
      "epoch": 0.6945145246161609,
      "grad_norm": 0.12925600404502335,
      "learning_rate": 4.508523003184085e-05,
      "loss": 1.0663,
      "step": 9635
    },
    {
      "epoch": 0.6945866070784978,
      "grad_norm": 0.11326974262704884,
      "learning_rate": 4.506571972344892e-05,
      "loss": 0.9966,
      "step": 9636
    },
    {
      "epoch": 0.6946586895408348,
      "grad_norm": 0.14198067896624367,
      "learning_rate": 4.504621240946668e-05,
      "loss": 0.9711,
      "step": 9637
    },
    {
      "epoch": 0.6947307720031717,
      "grad_norm": 0.12161570426186107,
      "learning_rate": 4.502670809095749e-05,
      "loss": 1.0562,
      "step": 9638
    },
    {
      "epoch": 0.6948028544655085,
      "grad_norm": 0.12257851870542777,
      "learning_rate": 4.500720676898452e-05,
      "loss": 1.0885,
      "step": 9639
    },
    {
      "epoch": 0.6948749369278454,
      "grad_norm": 0.11382863969493172,
      "learning_rate": 4.4987708444610763e-05,
      "loss": 0.9899,
      "step": 9640
    },
    {
      "epoch": 0.6949470193901823,
      "grad_norm": 0.1573137673580199,
      "learning_rate": 4.496821311889905e-05,
      "loss": 1.0726,
      "step": 9641
    },
    {
      "epoch": 0.6950191018525192,
      "grad_norm": 0.11212547880697715,
      "learning_rate": 4.494872079291205e-05,
      "loss": 1.0634,
      "step": 9642
    },
    {
      "epoch": 0.6950911843148562,
      "grad_norm": 0.11897658858648433,
      "learning_rate": 4.492923146771226e-05,
      "loss": 1.0561,
      "step": 9643
    },
    {
      "epoch": 0.6951632667771931,
      "grad_norm": 0.15986535884355604,
      "learning_rate": 4.4909745144362046e-05,
      "loss": 1.0359,
      "step": 9644
    },
    {
      "epoch": 0.69523534923953,
      "grad_norm": 0.13841971699631042,
      "learning_rate": 4.4890261823923564e-05,
      "loss": 1.0173,
      "step": 9645
    },
    {
      "epoch": 0.6953074317018669,
      "grad_norm": 0.14959546683307579,
      "learning_rate": 4.487078150745883e-05,
      "loss": 1.0428,
      "step": 9646
    },
    {
      "epoch": 0.6953795141642038,
      "grad_norm": 0.11748829174327927,
      "learning_rate": 4.485130419602972e-05,
      "loss": 1.04,
      "step": 9647
    },
    {
      "epoch": 0.6954515966265408,
      "grad_norm": 0.11843432113015415,
      "learning_rate": 4.483182989069791e-05,
      "loss": 1.0823,
      "step": 9648
    },
    {
      "epoch": 0.6955236790888777,
      "grad_norm": 0.14876749128952832,
      "learning_rate": 4.481235859252491e-05,
      "loss": 1.0627,
      "step": 9649
    },
    {
      "epoch": 0.6955957615512146,
      "grad_norm": 0.1405582941860378,
      "learning_rate": 4.479289030257209e-05,
      "loss": 1.0094,
      "step": 9650
    },
    {
      "epoch": 0.6956678440135515,
      "grad_norm": 0.11683117439021883,
      "learning_rate": 4.477342502190064e-05,
      "loss": 0.9948,
      "step": 9651
    },
    {
      "epoch": 0.6957399264758884,
      "grad_norm": 0.1435847387909175,
      "learning_rate": 4.47539627515716e-05,
      "loss": 1.0208,
      "step": 9652
    },
    {
      "epoch": 0.6958120089382254,
      "grad_norm": 0.11934510003903791,
      "learning_rate": 4.473450349264584e-05,
      "loss": 1.0134,
      "step": 9653
    },
    {
      "epoch": 0.6958840914005623,
      "grad_norm": 0.10643672746096078,
      "learning_rate": 4.471504724618405e-05,
      "loss": 1.0,
      "step": 9654
    },
    {
      "epoch": 0.6959561738628992,
      "grad_norm": 0.15260659441262295,
      "learning_rate": 4.469559401324679e-05,
      "loss": 1.0784,
      "step": 9655
    },
    {
      "epoch": 0.696028256325236,
      "grad_norm": 0.8548098708087872,
      "learning_rate": 4.467614379489442e-05,
      "loss": 0.9952,
      "step": 9656
    },
    {
      "epoch": 0.6961003387875729,
      "grad_norm": 0.1253159091480182,
      "learning_rate": 4.465669659218715e-05,
      "loss": 1.0485,
      "step": 9657
    },
    {
      "epoch": 0.6961724212499099,
      "grad_norm": 0.1358982129612408,
      "learning_rate": 4.463725240618504e-05,
      "loss": 1.0237,
      "step": 9658
    },
    {
      "epoch": 0.6962445037122468,
      "grad_norm": 0.11850995810686686,
      "learning_rate": 4.461781123794795e-05,
      "loss": 1.0388,
      "step": 9659
    },
    {
      "epoch": 0.6963165861745837,
      "grad_norm": 0.11829537913265911,
      "learning_rate": 4.459837308853561e-05,
      "loss": 1.0917,
      "step": 9660
    },
    {
      "epoch": 0.6963886686369206,
      "grad_norm": 0.14475516125701657,
      "learning_rate": 4.457893795900763e-05,
      "loss": 1.07,
      "step": 9661
    },
    {
      "epoch": 0.6964607510992575,
      "grad_norm": 0.11430390380916514,
      "learning_rate": 4.4559505850423254e-05,
      "loss": 1.0143,
      "step": 9662
    },
    {
      "epoch": 0.6965328335615945,
      "grad_norm": 0.11994598819824835,
      "learning_rate": 4.454007676384183e-05,
      "loss": 1.0282,
      "step": 9663
    },
    {
      "epoch": 0.6966049160239314,
      "grad_norm": 0.1221025708190463,
      "learning_rate": 4.452065070032239e-05,
      "loss": 1.0406,
      "step": 9664
    },
    {
      "epoch": 0.6966769984862683,
      "grad_norm": 0.12974313865623846,
      "learning_rate": 4.450122766092385e-05,
      "loss": 1.0241,
      "step": 9665
    },
    {
      "epoch": 0.6967490809486052,
      "grad_norm": 0.24887077707737806,
      "learning_rate": 4.4481807646704835e-05,
      "loss": 1.0409,
      "step": 9666
    },
    {
      "epoch": 0.6968211634109421,
      "grad_norm": 0.16053208376519038,
      "learning_rate": 4.446239065872402e-05,
      "loss": 1.0068,
      "step": 9667
    },
    {
      "epoch": 0.696893245873279,
      "grad_norm": 0.14294435256612564,
      "learning_rate": 4.444297669803981e-05,
      "loss": 1.0193,
      "step": 9668
    },
    {
      "epoch": 0.696965328335616,
      "grad_norm": 0.14583816305056152,
      "learning_rate": 4.442356576571035e-05,
      "loss": 1.037,
      "step": 9669
    },
    {
      "epoch": 0.6970374107979529,
      "grad_norm": 0.12477507465746544,
      "learning_rate": 4.4404157862793764e-05,
      "loss": 0.9909,
      "step": 9670
    },
    {
      "epoch": 0.6971094932602898,
      "grad_norm": 0.14146030740803978,
      "learning_rate": 4.438475299034791e-05,
      "loss": 1.0305,
      "step": 9671
    },
    {
      "epoch": 0.6971815757226267,
      "grad_norm": 0.11306648877604723,
      "learning_rate": 4.4365351149430645e-05,
      "loss": 1.1603,
      "step": 9672
    },
    {
      "epoch": 0.6972536581849635,
      "grad_norm": 0.1215718265386142,
      "learning_rate": 4.4345952341099416e-05,
      "loss": 0.9161,
      "step": 9673
    },
    {
      "epoch": 0.6973257406473005,
      "grad_norm": 0.13509358654837,
      "learning_rate": 4.4326556566411694e-05,
      "loss": 0.9817,
      "step": 9674
    },
    {
      "epoch": 0.6973978231096374,
      "grad_norm": 0.1189276116383918,
      "learning_rate": 4.4307163826424656e-05,
      "loss": 1.0599,
      "step": 9675
    },
    {
      "epoch": 0.6974699055719743,
      "grad_norm": 0.14574119093081855,
      "learning_rate": 4.4287774122195516e-05,
      "loss": 1.0441,
      "step": 9676
    },
    {
      "epoch": 0.6975419880343112,
      "grad_norm": 0.11805943333622435,
      "learning_rate": 4.4268387454781065e-05,
      "loss": 1.0493,
      "step": 9677
    },
    {
      "epoch": 0.6976140704966481,
      "grad_norm": 0.11894938132907786,
      "learning_rate": 4.424900382523809e-05,
      "loss": 0.9483,
      "step": 9678
    },
    {
      "epoch": 0.6976861529589851,
      "grad_norm": 0.16506880589841394,
      "learning_rate": 4.422962323462316e-05,
      "loss": 1.0003,
      "step": 9679
    },
    {
      "epoch": 0.697758235421322,
      "grad_norm": 0.18930149285134987,
      "learning_rate": 4.42102456839927e-05,
      "loss": 1.0991,
      "step": 9680
    },
    {
      "epoch": 0.6978303178836589,
      "grad_norm": 0.16046150519903213,
      "learning_rate": 4.419087117440297e-05,
      "loss": 1.0796,
      "step": 9681
    },
    {
      "epoch": 0.6979024003459958,
      "grad_norm": 0.12068857616451462,
      "learning_rate": 4.4171499706910034e-05,
      "loss": 1.0286,
      "step": 9682
    },
    {
      "epoch": 0.6979744828083327,
      "grad_norm": 0.11879292666453826,
      "learning_rate": 4.415213128256982e-05,
      "loss": 1.0528,
      "step": 9683
    },
    {
      "epoch": 0.6980465652706697,
      "grad_norm": 0.11168743819664008,
      "learning_rate": 4.413276590243809e-05,
      "loss": 1.0273,
      "step": 9684
    },
    {
      "epoch": 0.6981186477330066,
      "grad_norm": 0.1234780228120607,
      "learning_rate": 4.411340356757041e-05,
      "loss": 1.0119,
      "step": 9685
    },
    {
      "epoch": 0.6981907301953435,
      "grad_norm": 0.12220402593838221,
      "learning_rate": 4.4094044279022216e-05,
      "loss": 1.054,
      "step": 9686
    },
    {
      "epoch": 0.6982628126576804,
      "grad_norm": 0.12029643736120577,
      "learning_rate": 4.407468803784875e-05,
      "loss": 1.0166,
      "step": 9687
    },
    {
      "epoch": 0.6983348951200173,
      "grad_norm": 0.11275862074067339,
      "learning_rate": 4.4055334845105125e-05,
      "loss": 1.0129,
      "step": 9688
    },
    {
      "epoch": 0.6984069775823543,
      "grad_norm": 0.12438846246551326,
      "learning_rate": 4.4035984701846234e-05,
      "loss": 0.9838,
      "step": 9689
    },
    {
      "epoch": 0.6984790600446912,
      "grad_norm": 0.15388078650855208,
      "learning_rate": 4.4016637609126853e-05,
      "loss": 1.0136,
      "step": 9690
    },
    {
      "epoch": 0.698551142507028,
      "grad_norm": 0.11085705506625787,
      "learning_rate": 4.399729356800155e-05,
      "loss": 0.981,
      "step": 9691
    },
    {
      "epoch": 0.6986232249693649,
      "grad_norm": 0.13560604122134812,
      "learning_rate": 4.397795257952477e-05,
      "loss": 1.086,
      "step": 9692
    },
    {
      "epoch": 0.6986953074317018,
      "grad_norm": 0.12030353803778197,
      "learning_rate": 4.3958614644750774e-05,
      "loss": 0.9695,
      "step": 9693
    },
    {
      "epoch": 0.6987673898940387,
      "grad_norm": 0.1572239762620149,
      "learning_rate": 4.3939279764733624e-05,
      "loss": 1.0747,
      "step": 9694
    },
    {
      "epoch": 0.6988394723563757,
      "grad_norm": 0.13092805892926657,
      "learning_rate": 4.391994794052726e-05,
      "loss": 0.969,
      "step": 9695
    },
    {
      "epoch": 0.6989115548187126,
      "grad_norm": 0.11927800378205677,
      "learning_rate": 4.390061917318545e-05,
      "loss": 1.139,
      "step": 9696
    },
    {
      "epoch": 0.6989836372810495,
      "grad_norm": 0.12140225042378636,
      "learning_rate": 4.388129346376178e-05,
      "loss": 1.0156,
      "step": 9697
    },
    {
      "epoch": 0.6990557197433864,
      "grad_norm": 0.12329837093837363,
      "learning_rate": 4.386197081330965e-05,
      "loss": 1.0036,
      "step": 9698
    },
    {
      "epoch": 0.6991278022057233,
      "grad_norm": 0.12832052153703474,
      "learning_rate": 4.3842651222882346e-05,
      "loss": 0.9973,
      "step": 9699
    },
    {
      "epoch": 0.6991998846680603,
      "grad_norm": 0.6504372622545246,
      "learning_rate": 4.382333469353295e-05,
      "loss": 0.9588,
      "step": 9700
    },
    {
      "epoch": 0.6992719671303972,
      "grad_norm": 0.1374060933548918,
      "learning_rate": 4.380402122631441e-05,
      "loss": 0.9756,
      "step": 9701
    },
    {
      "epoch": 0.6993440495927341,
      "grad_norm": 0.1107876731774644,
      "learning_rate": 4.3784710822279394e-05,
      "loss": 1.0763,
      "step": 9702
    },
    {
      "epoch": 0.699416132055071,
      "grad_norm": 0.14191334354243615,
      "learning_rate": 4.376540348248058e-05,
      "loss": 1.0459,
      "step": 9703
    },
    {
      "epoch": 0.699488214517408,
      "grad_norm": 0.12036989760700746,
      "learning_rate": 4.374609920797038e-05,
      "loss": 0.957,
      "step": 9704
    },
    {
      "epoch": 0.6995602969797449,
      "grad_norm": 0.12646843396372623,
      "learning_rate": 4.372679799980106e-05,
      "loss": 1.0896,
      "step": 9705
    },
    {
      "epoch": 0.6996323794420818,
      "grad_norm": 0.11943489322354689,
      "learning_rate": 4.3707499859024615e-05,
      "loss": 1.0536,
      "step": 9706
    },
    {
      "epoch": 0.6997044619044187,
      "grad_norm": 0.23964519286462643,
      "learning_rate": 4.3688204786693075e-05,
      "loss": 1.0312,
      "step": 9707
    },
    {
      "epoch": 0.6997765443667555,
      "grad_norm": 0.13192143508523152,
      "learning_rate": 4.366891278385815e-05,
      "loss": 1.022,
      "step": 9708
    },
    {
      "epoch": 0.6998486268290924,
      "grad_norm": 0.10940087454236308,
      "learning_rate": 4.3649623851571464e-05,
      "loss": 0.9541,
      "step": 9709
    },
    {
      "epoch": 0.6999207092914294,
      "grad_norm": 0.10989995661334526,
      "learning_rate": 4.363033799088433e-05,
      "loss": 1.0125,
      "step": 9710
    },
    {
      "epoch": 0.6999927917537663,
      "grad_norm": 0.11769482477301069,
      "learning_rate": 4.361105520284812e-05,
      "loss": 0.9237,
      "step": 9711
    },
    {
      "epoch": 0.7000648742161032,
      "grad_norm": 0.15421946215347768,
      "learning_rate": 4.359177548851386e-05,
      "loss": 1.0217,
      "step": 9712
    },
    {
      "epoch": 0.7001369566784401,
      "grad_norm": 0.15437419308992703,
      "learning_rate": 4.357249884893252e-05,
      "loss": 1.0522,
      "step": 9713
    },
    {
      "epoch": 0.700209039140777,
      "grad_norm": 0.12104344025803351,
      "learning_rate": 4.355322528515476e-05,
      "loss": 1.0604,
      "step": 9714
    },
    {
      "epoch": 0.700281121603114,
      "grad_norm": 0.14013389534182916,
      "learning_rate": 4.353395479823119e-05,
      "loss": 1.0035,
      "step": 9715
    },
    {
      "epoch": 0.7003532040654509,
      "grad_norm": 0.13573856751028293,
      "learning_rate": 4.351468738921227e-05,
      "loss": 1.0704,
      "step": 9716
    },
    {
      "epoch": 0.7004252865277878,
      "grad_norm": 0.17139060360238437,
      "learning_rate": 4.349542305914826e-05,
      "loss": 1.0671,
      "step": 9717
    },
    {
      "epoch": 0.7004973689901247,
      "grad_norm": 0.12614878676028057,
      "learning_rate": 4.347616180908915e-05,
      "loss": 1.0199,
      "step": 9718
    },
    {
      "epoch": 0.7005694514524616,
      "grad_norm": 0.11823913026303723,
      "learning_rate": 4.345690364008488e-05,
      "loss": 0.9975,
      "step": 9719
    },
    {
      "epoch": 0.7006415339147986,
      "grad_norm": 0.13082944263969137,
      "learning_rate": 4.343764855318524e-05,
      "loss": 1.0,
      "step": 9720
    },
    {
      "epoch": 0.7007136163771355,
      "grad_norm": 0.9730894023578642,
      "learning_rate": 4.3418396549439814e-05,
      "loss": 1.0225,
      "step": 9721
    },
    {
      "epoch": 0.7007856988394724,
      "grad_norm": 0.12862456267000047,
      "learning_rate": 4.339914762989792e-05,
      "loss": 1.0679,
      "step": 9722
    },
    {
      "epoch": 0.7008577813018093,
      "grad_norm": 0.12173678201224958,
      "learning_rate": 4.3379901795608856e-05,
      "loss": 1.0007,
      "step": 9723
    },
    {
      "epoch": 0.7009298637641462,
      "grad_norm": 0.1131921949127812,
      "learning_rate": 4.336065904762164e-05,
      "loss": 1.0822,
      "step": 9724
    },
    {
      "epoch": 0.701001946226483,
      "grad_norm": 0.11248249451664843,
      "learning_rate": 4.3341419386985295e-05,
      "loss": 1.0117,
      "step": 9725
    },
    {
      "epoch": 0.70107402868882,
      "grad_norm": 0.2989574740773395,
      "learning_rate": 4.3322182814748436e-05,
      "loss": 1.0779,
      "step": 9726
    },
    {
      "epoch": 0.7011461111511569,
      "grad_norm": 0.10967463059799547,
      "learning_rate": 4.3302949331959675e-05,
      "loss": 1.0168,
      "step": 9727
    },
    {
      "epoch": 0.7012181936134938,
      "grad_norm": 0.12469199502904436,
      "learning_rate": 4.328371893966736e-05,
      "loss": 1.0895,
      "step": 9728
    },
    {
      "epoch": 0.7012902760758307,
      "grad_norm": 0.1188112045384966,
      "learning_rate": 4.3264491638919826e-05,
      "loss": 1.0466,
      "step": 9729
    },
    {
      "epoch": 0.7013623585381676,
      "grad_norm": 0.12708805267077547,
      "learning_rate": 4.324526743076504e-05,
      "loss": 0.9952,
      "step": 9730
    },
    {
      "epoch": 0.7014344410005046,
      "grad_norm": 0.16810315770995413,
      "learning_rate": 4.3226046316250927e-05,
      "loss": 1.0851,
      "step": 9731
    },
    {
      "epoch": 0.7015065234628415,
      "grad_norm": 0.12762312545384225,
      "learning_rate": 4.320682829642515e-05,
      "loss": 1.0519,
      "step": 9732
    },
    {
      "epoch": 0.7015786059251784,
      "grad_norm": 0.11567457737579315,
      "learning_rate": 4.31876133723354e-05,
      "loss": 1.0034,
      "step": 9733
    },
    {
      "epoch": 0.7016506883875153,
      "grad_norm": 0.12941910797877407,
      "learning_rate": 4.316840154502895e-05,
      "loss": 1.0243,
      "step": 9734
    },
    {
      "epoch": 0.7017227708498522,
      "grad_norm": 0.1973858763910186,
      "learning_rate": 4.3149192815553044e-05,
      "loss": 1.0657,
      "step": 9735
    },
    {
      "epoch": 0.7017948533121892,
      "grad_norm": 0.1354048154882434,
      "learning_rate": 4.3129987184954726e-05,
      "loss": 1.0169,
      "step": 9736
    },
    {
      "epoch": 0.7018669357745261,
      "grad_norm": 0.1351296770293007,
      "learning_rate": 4.311078465428088e-05,
      "loss": 0.986,
      "step": 9737
    },
    {
      "epoch": 0.701939018236863,
      "grad_norm": 0.12145529296173091,
      "learning_rate": 4.3091585224578214e-05,
      "loss": 0.9944,
      "step": 9738
    },
    {
      "epoch": 0.7020111006991999,
      "grad_norm": 0.24776837598587592,
      "learning_rate": 4.307238889689327e-05,
      "loss": 1.023,
      "step": 9739
    },
    {
      "epoch": 0.7020831831615368,
      "grad_norm": 2.2958951810781927,
      "learning_rate": 4.305319567227243e-05,
      "loss": 1.0193,
      "step": 9740
    },
    {
      "epoch": 0.7021552656238738,
      "grad_norm": 0.5249376022427231,
      "learning_rate": 4.303400555176187e-05,
      "loss": 1.0674,
      "step": 9741
    },
    {
      "epoch": 0.7022273480862107,
      "grad_norm": 0.1319105136151909,
      "learning_rate": 4.301481853640765e-05,
      "loss": 1.046,
      "step": 9742
    },
    {
      "epoch": 0.7022994305485475,
      "grad_norm": 0.12399626878698637,
      "learning_rate": 4.299563462725563e-05,
      "loss": 1.0186,
      "step": 9743
    },
    {
      "epoch": 0.7023715130108844,
      "grad_norm": 0.126201817105347,
      "learning_rate": 4.2976453825351495e-05,
      "loss": 1.1042,
      "step": 9744
    },
    {
      "epoch": 0.7024435954732213,
      "grad_norm": 0.1868415515273449,
      "learning_rate": 4.2957276131740766e-05,
      "loss": 1.0677,
      "step": 9745
    },
    {
      "epoch": 0.7025156779355582,
      "grad_norm": 0.11971284539860798,
      "learning_rate": 4.293810154746881e-05,
      "loss": 0.9977,
      "step": 9746
    },
    {
      "epoch": 0.7025877603978952,
      "grad_norm": 0.11651913627230881,
      "learning_rate": 4.291893007358081e-05,
      "loss": 1.0348,
      "step": 9747
    },
    {
      "epoch": 0.7026598428602321,
      "grad_norm": 0.13624472413365077,
      "learning_rate": 4.2899761711121787e-05,
      "loss": 1.0772,
      "step": 9748
    },
    {
      "epoch": 0.702731925322569,
      "grad_norm": 0.13869587244535067,
      "learning_rate": 4.288059646113658e-05,
      "loss": 0.9635,
      "step": 9749
    },
    {
      "epoch": 0.7028040077849059,
      "grad_norm": 0.14479492238938704,
      "learning_rate": 4.286143432466987e-05,
      "loss": 0.9845,
      "step": 9750
    },
    {
      "epoch": 0.7028760902472428,
      "grad_norm": 0.22847679358381906,
      "learning_rate": 4.284227530276617e-05,
      "loss": 1.0003,
      "step": 9751
    },
    {
      "epoch": 0.7029481727095798,
      "grad_norm": 0.1449481793023658,
      "learning_rate": 4.28231193964698e-05,
      "loss": 1.1101,
      "step": 9752
    },
    {
      "epoch": 0.7030202551719167,
      "grad_norm": 0.12162791040807217,
      "learning_rate": 4.280396660682497e-05,
      "loss": 1.0253,
      "step": 9753
    },
    {
      "epoch": 0.7030923376342536,
      "grad_norm": 0.12732561479602675,
      "learning_rate": 4.2784816934875625e-05,
      "loss": 1.0822,
      "step": 9754
    },
    {
      "epoch": 0.7031644200965905,
      "grad_norm": 0.16631110599592086,
      "learning_rate": 4.276567038166563e-05,
      "loss": 0.9607,
      "step": 9755
    },
    {
      "epoch": 0.7032365025589274,
      "grad_norm": 0.11079720155055431,
      "learning_rate": 4.274652694823862e-05,
      "loss": 0.9829,
      "step": 9756
    },
    {
      "epoch": 0.7033085850212644,
      "grad_norm": 0.1492916546402305,
      "learning_rate": 4.272738663563811e-05,
      "loss": 0.9993,
      "step": 9757
    },
    {
      "epoch": 0.7033806674836013,
      "grad_norm": 0.10512672072823509,
      "learning_rate": 4.2708249444907436e-05,
      "loss": 1.0658,
      "step": 9758
    },
    {
      "epoch": 0.7034527499459382,
      "grad_norm": 0.11891183407396329,
      "learning_rate": 4.268911537708964e-05,
      "loss": 1.0464,
      "step": 9759
    },
    {
      "epoch": 0.703524832408275,
      "grad_norm": 0.1256293532720422,
      "learning_rate": 4.266998443322781e-05,
      "loss": 0.989,
      "step": 9760
    },
    {
      "epoch": 0.7035969148706119,
      "grad_norm": 0.13353761378765086,
      "learning_rate": 4.265085661436471e-05,
      "loss": 0.9931,
      "step": 9761
    },
    {
      "epoch": 0.7036689973329489,
      "grad_norm": 0.1216699089909593,
      "learning_rate": 4.2631731921543026e-05,
      "loss": 1.0421,
      "step": 9762
    },
    {
      "epoch": 0.7037410797952858,
      "grad_norm": 0.13028221796248682,
      "learning_rate": 4.2612610355805114e-05,
      "loss": 1.1042,
      "step": 9763
    },
    {
      "epoch": 0.7038131622576227,
      "grad_norm": 0.13203160688326107,
      "learning_rate": 4.259349191819337e-05,
      "loss": 0.9961,
      "step": 9764
    },
    {
      "epoch": 0.7038852447199596,
      "grad_norm": 0.13216977195416066,
      "learning_rate": 4.25743766097499e-05,
      "loss": 0.9808,
      "step": 9765
    },
    {
      "epoch": 0.7039573271822965,
      "grad_norm": 0.17539793801638828,
      "learning_rate": 4.255526443151668e-05,
      "loss": 1.0259,
      "step": 9766
    },
    {
      "epoch": 0.7040294096446335,
      "grad_norm": 0.14299514425214438,
      "learning_rate": 4.2536155384535426e-05,
      "loss": 1.1229,
      "step": 9767
    },
    {
      "epoch": 0.7041014921069704,
      "grad_norm": 0.10920858149393685,
      "learning_rate": 4.251704946984776e-05,
      "loss": 0.9856,
      "step": 9768
    },
    {
      "epoch": 0.7041735745693073,
      "grad_norm": 0.20021819237853813,
      "learning_rate": 4.2497946688495184e-05,
      "loss": 1.0764,
      "step": 9769
    },
    {
      "epoch": 0.7042456570316442,
      "grad_norm": 0.13000070334932237,
      "learning_rate": 4.247884704151898e-05,
      "loss": 1.0004,
      "step": 9770
    },
    {
      "epoch": 0.7043177394939811,
      "grad_norm": 0.13863211441365905,
      "learning_rate": 4.2459750529960176e-05,
      "loss": 1.0641,
      "step": 9771
    },
    {
      "epoch": 0.704389821956318,
      "grad_norm": 0.11535159369215756,
      "learning_rate": 4.24406571548597e-05,
      "loss": 1.0034,
      "step": 9772
    },
    {
      "epoch": 0.704461904418655,
      "grad_norm": 0.10169774629650936,
      "learning_rate": 4.242156691725838e-05,
      "loss": 0.9532,
      "step": 9773
    },
    {
      "epoch": 0.7045339868809919,
      "grad_norm": 0.12816413202129237,
      "learning_rate": 4.240247981819683e-05,
      "loss": 1.0711,
      "step": 9774
    },
    {
      "epoch": 0.7046060693433288,
      "grad_norm": 0.12579494743083308,
      "learning_rate": 4.238339585871536e-05,
      "loss": 0.9645,
      "step": 9775
    },
    {
      "epoch": 0.7046781518056657,
      "grad_norm": 0.1161246406393063,
      "learning_rate": 4.236431503985423e-05,
      "loss": 1.0405,
      "step": 9776
    },
    {
      "epoch": 0.7047502342680025,
      "grad_norm": 0.18940952510809453,
      "learning_rate": 4.23452373626536e-05,
      "loss": 1.0446,
      "step": 9777
    },
    {
      "epoch": 0.7048223167303395,
      "grad_norm": 0.11829261553335048,
      "learning_rate": 4.232616282815336e-05,
      "loss": 1.0031,
      "step": 9778
    },
    {
      "epoch": 0.7048943991926764,
      "grad_norm": 0.11803988379321145,
      "learning_rate": 4.230709143739317e-05,
      "loss": 1.0494,
      "step": 9779
    },
    {
      "epoch": 0.7049664816550133,
      "grad_norm": 0.1273949090336124,
      "learning_rate": 4.228802319141264e-05,
      "loss": 1.0159,
      "step": 9780
    },
    {
      "epoch": 0.7050385641173502,
      "grad_norm": 0.12408910439667825,
      "learning_rate": 4.2268958091251107e-05,
      "loss": 1.078,
      "step": 9781
    },
    {
      "epoch": 0.7051106465796871,
      "grad_norm": 0.1181990195992283,
      "learning_rate": 4.224989613794792e-05,
      "loss": 1.0032,
      "step": 9782
    },
    {
      "epoch": 0.7051827290420241,
      "grad_norm": 0.15882900938071243,
      "learning_rate": 4.2230837332541994e-05,
      "loss": 1.0359,
      "step": 9783
    },
    {
      "epoch": 0.705254811504361,
      "grad_norm": 0.18290090692885855,
      "learning_rate": 4.221178167607226e-05,
      "loss": 0.9937,
      "step": 9784
    },
    {
      "epoch": 0.7053268939666979,
      "grad_norm": 0.11541643332386743,
      "learning_rate": 4.219272916957736e-05,
      "loss": 1.0165,
      "step": 9785
    },
    {
      "epoch": 0.7053989764290348,
      "grad_norm": 0.11677009243231834,
      "learning_rate": 4.2173679814095964e-05,
      "loss": 0.9683,
      "step": 9786
    },
    {
      "epoch": 0.7054710588913717,
      "grad_norm": 0.113381864176993,
      "learning_rate": 4.215463361066632e-05,
      "loss": 1.0333,
      "step": 9787
    },
    {
      "epoch": 0.7055431413537087,
      "grad_norm": 0.11623214911506624,
      "learning_rate": 4.213559056032662e-05,
      "loss": 1.2167,
      "step": 9788
    },
    {
      "epoch": 0.7056152238160456,
      "grad_norm": 0.6015711577220871,
      "learning_rate": 4.211655066411492e-05,
      "loss": 1.0518,
      "step": 9789
    },
    {
      "epoch": 0.7056873062783825,
      "grad_norm": 0.11848895248009761,
      "learning_rate": 4.2097513923069045e-05,
      "loss": 0.9874,
      "step": 9790
    },
    {
      "epoch": 0.7057593887407194,
      "grad_norm": 0.11500083253444567,
      "learning_rate": 4.207848033822668e-05,
      "loss": 1.0912,
      "step": 9791
    },
    {
      "epoch": 0.7058314712030563,
      "grad_norm": 0.10994680204750104,
      "learning_rate": 4.20594499106253e-05,
      "loss": 1.0252,
      "step": 9792
    },
    {
      "epoch": 0.7059035536653933,
      "grad_norm": 0.12004598397091466,
      "learning_rate": 4.204042264130227e-05,
      "loss": 1.0787,
      "step": 9793
    },
    {
      "epoch": 0.7059756361277302,
      "grad_norm": 0.13473575852838202,
      "learning_rate": 4.2021398531294706e-05,
      "loss": 1.0712,
      "step": 9794
    },
    {
      "epoch": 0.706047718590067,
      "grad_norm": 0.11200152479066229,
      "learning_rate": 4.200237758163963e-05,
      "loss": 1.015,
      "step": 9795
    },
    {
      "epoch": 0.7061198010524039,
      "grad_norm": 0.13787149328807013,
      "learning_rate": 4.198335979337383e-05,
      "loss": 1.0304,
      "step": 9796
    },
    {
      "epoch": 0.7061918835147408,
      "grad_norm": 0.12473409740220705,
      "learning_rate": 4.196434516753396e-05,
      "loss": 1.0206,
      "step": 9797
    },
    {
      "epoch": 0.7062639659770777,
      "grad_norm": 0.14669105887766448,
      "learning_rate": 4.194533370515648e-05,
      "loss": 1.0115,
      "step": 9798
    },
    {
      "epoch": 0.7063360484394147,
      "grad_norm": 0.14134376636437287,
      "learning_rate": 4.192632540727768e-05,
      "loss": 0.9831,
      "step": 9799
    },
    {
      "epoch": 0.7064081309017516,
      "grad_norm": 0.12893851679685772,
      "learning_rate": 4.190732027493368e-05,
      "loss": 0.9849,
      "step": 9800
    },
    {
      "epoch": 0.7064802133640885,
      "grad_norm": 0.11785778888106965,
      "learning_rate": 4.1888318309160445e-05,
      "loss": 1.0108,
      "step": 9801
    },
    {
      "epoch": 0.7065522958264254,
      "grad_norm": 0.16090674480423542,
      "learning_rate": 4.1869319510993755e-05,
      "loss": 0.9873,
      "step": 9802
    },
    {
      "epoch": 0.7066243782887623,
      "grad_norm": 0.11715192882032738,
      "learning_rate": 4.185032388146919e-05,
      "loss": 1.0357,
      "step": 9803
    },
    {
      "epoch": 0.7066964607510993,
      "grad_norm": 0.11715773024944397,
      "learning_rate": 4.18313314216222e-05,
      "loss": 1.0103,
      "step": 9804
    },
    {
      "epoch": 0.7067685432134362,
      "grad_norm": 0.11771317190164836,
      "learning_rate": 4.1812342132488036e-05,
      "loss": 1.0532,
      "step": 9805
    },
    {
      "epoch": 0.7068406256757731,
      "grad_norm": 0.21364535816422955,
      "learning_rate": 4.179335601510178e-05,
      "loss": 1.0345,
      "step": 9806
    },
    {
      "epoch": 0.70691270813811,
      "grad_norm": 0.15640414492496582,
      "learning_rate": 4.177437307049836e-05,
      "loss": 1.0173,
      "step": 9807
    },
    {
      "epoch": 0.7069847906004469,
      "grad_norm": 0.11328843443617256,
      "learning_rate": 4.175539329971252e-05,
      "loss": 1.0137,
      "step": 9808
    },
    {
      "epoch": 0.7070568730627839,
      "grad_norm": 0.1624504725114403,
      "learning_rate": 4.1736416703778804e-05,
      "loss": 1.0521,
      "step": 9809
    },
    {
      "epoch": 0.7071289555251208,
      "grad_norm": 0.12120667670158897,
      "learning_rate": 4.1717443283731615e-05,
      "loss": 1.0064,
      "step": 9810
    },
    {
      "epoch": 0.7072010379874577,
      "grad_norm": 0.11423675416117597,
      "learning_rate": 4.169847304060522e-05,
      "loss": 1.0555,
      "step": 9811
    },
    {
      "epoch": 0.7072731204497945,
      "grad_norm": 0.11263750881329143,
      "learning_rate": 4.167950597543356e-05,
      "loss": 1.0231,
      "step": 9812
    },
    {
      "epoch": 0.7073452029121314,
      "grad_norm": 0.11622395652036539,
      "learning_rate": 4.16605420892506e-05,
      "loss": 1.0394,
      "step": 9813
    },
    {
      "epoch": 0.7074172853744684,
      "grad_norm": 0.1045209610957669,
      "learning_rate": 4.164158138309002e-05,
      "loss": 1.0377,
      "step": 9814
    },
    {
      "epoch": 0.7074893678368053,
      "grad_norm": 0.11686226833493119,
      "learning_rate": 4.1622623857985374e-05,
      "loss": 1.0225,
      "step": 9815
    },
    {
      "epoch": 0.7075614502991422,
      "grad_norm": 0.12167173174621966,
      "learning_rate": 4.160366951496992e-05,
      "loss": 1.0535,
      "step": 9816
    },
    {
      "epoch": 0.7076335327614791,
      "grad_norm": 0.11916932684164881,
      "learning_rate": 4.158471835507695e-05,
      "loss": 0.9846,
      "step": 9817
    },
    {
      "epoch": 0.707705615223816,
      "grad_norm": 0.10751707162265454,
      "learning_rate": 4.156577037933942e-05,
      "loss": 1.0547,
      "step": 9818
    },
    {
      "epoch": 0.707777697686153,
      "grad_norm": 0.127970980286589,
      "learning_rate": 4.154682558879021e-05,
      "loss": 1.0312,
      "step": 9819
    },
    {
      "epoch": 0.7078497801484899,
      "grad_norm": 0.11495015514184594,
      "learning_rate": 4.152788398446186e-05,
      "loss": 1.0835,
      "step": 9820
    },
    {
      "epoch": 0.7079218626108268,
      "grad_norm": 0.11430303206471275,
      "learning_rate": 4.150894556738698e-05,
      "loss": 0.9914,
      "step": 9821
    },
    {
      "epoch": 0.7079939450731637,
      "grad_norm": 0.11979770084823677,
      "learning_rate": 4.149001033859788e-05,
      "loss": 1.0881,
      "step": 9822
    },
    {
      "epoch": 0.7080660275355006,
      "grad_norm": 0.1311386654583346,
      "learning_rate": 4.147107829912663e-05,
      "loss": 0.9996,
      "step": 9823
    },
    {
      "epoch": 0.7081381099978376,
      "grad_norm": 0.12306244904418247,
      "learning_rate": 4.145214945000523e-05,
      "loss": 0.9891,
      "step": 9824
    },
    {
      "epoch": 0.7082101924601745,
      "grad_norm": 0.12949569903881755,
      "learning_rate": 4.143322379226542e-05,
      "loss": 1.0063,
      "step": 9825
    },
    {
      "epoch": 0.7082822749225114,
      "grad_norm": 0.13996894998676138,
      "learning_rate": 4.141430132693895e-05,
      "loss": 1.0649,
      "step": 9826
    },
    {
      "epoch": 0.7083543573848483,
      "grad_norm": 0.12983310091404965,
      "learning_rate": 4.139538205505714e-05,
      "loss": 1.0631,
      "step": 9827
    },
    {
      "epoch": 0.7084264398471852,
      "grad_norm": 0.27697546079321855,
      "learning_rate": 4.13764659776513e-05,
      "loss": 0.9937,
      "step": 9828
    },
    {
      "epoch": 0.708498522309522,
      "grad_norm": 0.12355425182589914,
      "learning_rate": 4.135755309575249e-05,
      "loss": 1.064,
      "step": 9829
    },
    {
      "epoch": 0.708570604771859,
      "grad_norm": 0.12302662354128382,
      "learning_rate": 4.133864341039174e-05,
      "loss": 1.0398,
      "step": 9830
    },
    {
      "epoch": 0.7086426872341959,
      "grad_norm": 0.5644729840866681,
      "learning_rate": 4.1319736922599686e-05,
      "loss": 1.0826,
      "step": 9831
    },
    {
      "epoch": 0.7087147696965328,
      "grad_norm": 0.15014927960699787,
      "learning_rate": 4.1300833633406935e-05,
      "loss": 1.0755,
      "step": 9832
    },
    {
      "epoch": 0.7087868521588697,
      "grad_norm": 0.1681038364195877,
      "learning_rate": 4.12819335438439e-05,
      "loss": 1.0044,
      "step": 9833
    },
    {
      "epoch": 0.7088589346212066,
      "grad_norm": 0.1238956347525188,
      "learning_rate": 4.1263036654940775e-05,
      "loss": 0.9811,
      "step": 9834
    },
    {
      "epoch": 0.7089310170835436,
      "grad_norm": 0.11618656611739009,
      "learning_rate": 4.1244142967727636e-05,
      "loss": 0.9834,
      "step": 9835
    },
    {
      "epoch": 0.7090030995458805,
      "grad_norm": 0.14166076897255478,
      "learning_rate": 4.122525248323436e-05,
      "loss": 1.0287,
      "step": 9836
    },
    {
      "epoch": 0.7090751820082174,
      "grad_norm": 0.10863117180156262,
      "learning_rate": 4.120636520249063e-05,
      "loss": 1.1016,
      "step": 9837
    },
    {
      "epoch": 0.7091472644705543,
      "grad_norm": 0.11516367801307123,
      "learning_rate": 4.1187481126525987e-05,
      "loss": 1.083,
      "step": 9838
    },
    {
      "epoch": 0.7092193469328912,
      "grad_norm": 0.12359883545131783,
      "learning_rate": 4.116860025636976e-05,
      "loss": 1.0645,
      "step": 9839
    },
    {
      "epoch": 0.7092914293952282,
      "grad_norm": 0.11311666769666823,
      "learning_rate": 4.114972259305115e-05,
      "loss": 1.0458,
      "step": 9840
    },
    {
      "epoch": 0.7093635118575651,
      "grad_norm": 0.6981236786972612,
      "learning_rate": 4.113084813759914e-05,
      "loss": 1.1338,
      "step": 9841
    },
    {
      "epoch": 0.709435594319902,
      "grad_norm": 0.11977471643243372,
      "learning_rate": 4.111197689104258e-05,
      "loss": 1.0385,
      "step": 9842
    },
    {
      "epoch": 0.7095076767822389,
      "grad_norm": 0.12704922278764996,
      "learning_rate": 4.109310885441009e-05,
      "loss": 0.9863,
      "step": 9843
    },
    {
      "epoch": 0.7095797592445758,
      "grad_norm": 0.1127062853635952,
      "learning_rate": 4.1074244028730166e-05,
      "loss": 1.0175,
      "step": 9844
    },
    {
      "epoch": 0.7096518417069128,
      "grad_norm": 0.13995027443762884,
      "learning_rate": 4.105538241503112e-05,
      "loss": 1.0811,
      "step": 9845
    },
    {
      "epoch": 0.7097239241692497,
      "grad_norm": 0.12538527586877135,
      "learning_rate": 4.103652401434106e-05,
      "loss": 1.1195,
      "step": 9846
    },
    {
      "epoch": 0.7097960066315865,
      "grad_norm": 0.11913414590863025,
      "learning_rate": 4.101766882768794e-05,
      "loss": 0.9616,
      "step": 9847
    },
    {
      "epoch": 0.7098680890939234,
      "grad_norm": 0.11834773294918657,
      "learning_rate": 4.099881685609954e-05,
      "loss": 1.0595,
      "step": 9848
    },
    {
      "epoch": 0.7099401715562603,
      "grad_norm": 0.11897984376637573,
      "learning_rate": 4.097996810060346e-05,
      "loss": 1.1176,
      "step": 9849
    },
    {
      "epoch": 0.7100122540185972,
      "grad_norm": 0.11403890399535359,
      "learning_rate": 4.096112256222713e-05,
      "loss": 1.0924,
      "step": 9850
    },
    {
      "epoch": 0.7100843364809342,
      "grad_norm": 0.14214249725777253,
      "learning_rate": 4.09422802419978e-05,
      "loss": 1.0633,
      "step": 9851
    },
    {
      "epoch": 0.7101564189432711,
      "grad_norm": 0.11865510697429571,
      "learning_rate": 4.092344114094254e-05,
      "loss": 1.0245,
      "step": 9852
    },
    {
      "epoch": 0.710228501405608,
      "grad_norm": 0.34792583920510367,
      "learning_rate": 4.0904605260088244e-05,
      "loss": 0.9676,
      "step": 9853
    },
    {
      "epoch": 0.7103005838679449,
      "grad_norm": 0.12833450225408,
      "learning_rate": 4.088577260046166e-05,
      "loss": 1.0174,
      "step": 9854
    },
    {
      "epoch": 0.7103726663302818,
      "grad_norm": 0.12426120832431746,
      "learning_rate": 4.086694316308934e-05,
      "loss": 1.0724,
      "step": 9855
    },
    {
      "epoch": 0.7104447487926188,
      "grad_norm": 0.12260280416901657,
      "learning_rate": 4.084811694899757e-05,
      "loss": 1.0356,
      "step": 9856
    },
    {
      "epoch": 0.7105168312549557,
      "grad_norm": 0.1297566491140655,
      "learning_rate": 4.0829293959212656e-05,
      "loss": 1.0118,
      "step": 9857
    },
    {
      "epoch": 0.7105889137172926,
      "grad_norm": 0.12350393228216398,
      "learning_rate": 4.081047419476057e-05,
      "loss": 0.9636,
      "step": 9858
    },
    {
      "epoch": 0.7106609961796295,
      "grad_norm": 0.1785649824235982,
      "learning_rate": 4.0791657656667205e-05,
      "loss": 1.0543,
      "step": 9859
    },
    {
      "epoch": 0.7107330786419664,
      "grad_norm": 0.13917693365807488,
      "learning_rate": 4.077284434595813e-05,
      "loss": 1.022,
      "step": 9860
    },
    {
      "epoch": 0.7108051611043034,
      "grad_norm": 0.12991063473781503,
      "learning_rate": 4.0754034263658916e-05,
      "loss": 1.0057,
      "step": 9861
    },
    {
      "epoch": 0.7108772435666403,
      "grad_norm": 0.15827171143837843,
      "learning_rate": 4.0735227410794875e-05,
      "loss": 1.0222,
      "step": 9862
    },
    {
      "epoch": 0.7109493260289772,
      "grad_norm": 0.11712501381595898,
      "learning_rate": 4.071642378839117e-05,
      "loss": 1.0721,
      "step": 9863
    },
    {
      "epoch": 0.711021408491314,
      "grad_norm": 0.13937916196531933,
      "learning_rate": 4.069762339747266e-05,
      "loss": 0.9857,
      "step": 9864
    },
    {
      "epoch": 0.7110934909536509,
      "grad_norm": 0.1409350688105421,
      "learning_rate": 4.067882623906425e-05,
      "loss": 1.0895,
      "step": 9865
    },
    {
      "epoch": 0.7111655734159879,
      "grad_norm": 0.11492291608027776,
      "learning_rate": 4.066003231419051e-05,
      "loss": 1.0103,
      "step": 9866
    },
    {
      "epoch": 0.7112376558783248,
      "grad_norm": 0.19266063296236363,
      "learning_rate": 4.0641241623875916e-05,
      "loss": 1.0193,
      "step": 9867
    },
    {
      "epoch": 0.7113097383406617,
      "grad_norm": 0.11099153958038507,
      "learning_rate": 4.062245416914465e-05,
      "loss": 1.0853,
      "step": 9868
    },
    {
      "epoch": 0.7113818208029986,
      "grad_norm": 0.11389645878726984,
      "learning_rate": 4.060366995102082e-05,
      "loss": 1.0526,
      "step": 9869
    },
    {
      "epoch": 0.7114539032653355,
      "grad_norm": 0.11478118985508734,
      "learning_rate": 4.0584888970528364e-05,
      "loss": 1.0057,
      "step": 9870
    },
    {
      "epoch": 0.7115259857276724,
      "grad_norm": 0.11133461144910917,
      "learning_rate": 4.0566111228691064e-05,
      "loss": 1.0328,
      "step": 9871
    },
    {
      "epoch": 0.7115980681900094,
      "grad_norm": 1.3797176737834342,
      "learning_rate": 4.054733672653236e-05,
      "loss": 1.0368,
      "step": 9872
    },
    {
      "epoch": 0.7116701506523463,
      "grad_norm": 0.12177727394341296,
      "learning_rate": 4.052856546507565e-05,
      "loss": 1.0158,
      "step": 9873
    },
    {
      "epoch": 0.7117422331146832,
      "grad_norm": 0.12597811079743232,
      "learning_rate": 4.050979744534421e-05,
      "loss": 1.1157,
      "step": 9874
    },
    {
      "epoch": 0.7118143155770201,
      "grad_norm": 0.10237478547655414,
      "learning_rate": 4.0491032668361054e-05,
      "loss": 1.0586,
      "step": 9875
    },
    {
      "epoch": 0.711886398039357,
      "grad_norm": 0.12331599626342206,
      "learning_rate": 4.047227113514897e-05,
      "loss": 0.9761,
      "step": 9876
    },
    {
      "epoch": 0.711958480501694,
      "grad_norm": 0.11235606321391052,
      "learning_rate": 4.0453512846730646e-05,
      "loss": 0.9759,
      "step": 9877
    },
    {
      "epoch": 0.7120305629640309,
      "grad_norm": 0.12865587233669024,
      "learning_rate": 4.0434757804128564e-05,
      "loss": 1.0164,
      "step": 9878
    },
    {
      "epoch": 0.7121026454263678,
      "grad_norm": 0.13051800929780377,
      "learning_rate": 4.041600600836514e-05,
      "loss": 1.0264,
      "step": 9879
    },
    {
      "epoch": 0.7121747278887047,
      "grad_norm": 0.11876309928180388,
      "learning_rate": 4.03972574604624e-05,
      "loss": 1.0129,
      "step": 9880
    },
    {
      "epoch": 0.7122468103510415,
      "grad_norm": 0.11466209153433181,
      "learning_rate": 4.037851216144236e-05,
      "loss": 1.0144,
      "step": 9881
    },
    {
      "epoch": 0.7123188928133785,
      "grad_norm": 0.15095837879029522,
      "learning_rate": 4.035977011232676e-05,
      "loss": 1.0259,
      "step": 9882
    },
    {
      "epoch": 0.7123909752757154,
      "grad_norm": 0.11807812218313729,
      "learning_rate": 4.0341031314137304e-05,
      "loss": 1.0657,
      "step": 9883
    },
    {
      "epoch": 0.7124630577380523,
      "grad_norm": 0.12180522422160651,
      "learning_rate": 4.0322295767895335e-05,
      "loss": 1.0963,
      "step": 9884
    },
    {
      "epoch": 0.7125351402003892,
      "grad_norm": 0.11633056221115401,
      "learning_rate": 4.030356347462214e-05,
      "loss": 1.0586,
      "step": 9885
    },
    {
      "epoch": 0.7126072226627261,
      "grad_norm": 0.1391057225633929,
      "learning_rate": 4.028483443533877e-05,
      "loss": 1.0859,
      "step": 9886
    },
    {
      "epoch": 0.7126793051250631,
      "grad_norm": 0.11015555783851987,
      "learning_rate": 4.0266108651066216e-05,
      "loss": 1.006,
      "step": 9887
    },
    {
      "epoch": 0.7127513875874,
      "grad_norm": 0.3761453457412791,
      "learning_rate": 4.0247386122825095e-05,
      "loss": 0.9915,
      "step": 9888
    },
    {
      "epoch": 0.7128234700497369,
      "grad_norm": 0.15689538753044513,
      "learning_rate": 4.0228666851635985e-05,
      "loss": 0.9352,
      "step": 9889
    },
    {
      "epoch": 0.7128955525120738,
      "grad_norm": 0.12932043687893344,
      "learning_rate": 4.020995083851927e-05,
      "loss": 1.0369,
      "step": 9890
    },
    {
      "epoch": 0.7129676349744107,
      "grad_norm": 0.13105885639782175,
      "learning_rate": 4.019123808449514e-05,
      "loss": 1.0373,
      "step": 9891
    },
    {
      "epoch": 0.7130397174367477,
      "grad_norm": 0.6821600234761491,
      "learning_rate": 4.017252859058359e-05,
      "loss": 1.0989,
      "step": 9892
    },
    {
      "epoch": 0.7131117998990846,
      "grad_norm": 0.13627629525841284,
      "learning_rate": 4.015382235780446e-05,
      "loss": 1.0263,
      "step": 9893
    },
    {
      "epoch": 0.7131838823614215,
      "grad_norm": 0.11363110611526636,
      "learning_rate": 4.0135119387177425e-05,
      "loss": 1.0181,
      "step": 9894
    },
    {
      "epoch": 0.7132559648237584,
      "grad_norm": 0.18598006195113093,
      "learning_rate": 4.011641967972193e-05,
      "loss": 0.9395,
      "step": 9895
    },
    {
      "epoch": 0.7133280472860953,
      "grad_norm": 0.1484758871798772,
      "learning_rate": 4.009772323645731e-05,
      "loss": 1.0499,
      "step": 9896
    },
    {
      "epoch": 0.7134001297484323,
      "grad_norm": 0.11028984196041296,
      "learning_rate": 4.007903005840267e-05,
      "loss": 1.0717,
      "step": 9897
    },
    {
      "epoch": 0.7134722122107692,
      "grad_norm": 0.12075954665761181,
      "learning_rate": 4.006034014657696e-05,
      "loss": 1.0584,
      "step": 9898
    },
    {
      "epoch": 0.713544294673106,
      "grad_norm": 0.309783565635684,
      "learning_rate": 4.0041653501998946e-05,
      "loss": 0.8992,
      "step": 9899
    },
    {
      "epoch": 0.7136163771354429,
      "grad_norm": 0.11985839993977716,
      "learning_rate": 4.002297012568722e-05,
      "loss": 1.0118,
      "step": 9900
    },
    {
      "epoch": 0.7136884595977798,
      "grad_norm": 0.1254963515918048,
      "learning_rate": 4.0004290018660195e-05,
      "loss": 1.057,
      "step": 9901
    },
    {
      "epoch": 0.7137605420601167,
      "grad_norm": 0.13546766963203163,
      "learning_rate": 3.99856131819361e-05,
      "loss": 1.0622,
      "step": 9902
    },
    {
      "epoch": 0.7138326245224537,
      "grad_norm": 0.11053021552380493,
      "learning_rate": 3.9966939616532994e-05,
      "loss": 0.9283,
      "step": 9903
    },
    {
      "epoch": 0.7139047069847906,
      "grad_norm": 0.12727135558082867,
      "learning_rate": 3.9948269323468744e-05,
      "loss": 0.9964,
      "step": 9904
    },
    {
      "epoch": 0.7139767894471275,
      "grad_norm": 0.11288207972223598,
      "learning_rate": 3.992960230376105e-05,
      "loss": 1.0323,
      "step": 9905
    },
    {
      "epoch": 0.7140488719094644,
      "grad_norm": 0.17059656746102983,
      "learning_rate": 3.9910938558427445e-05,
      "loss": 1.0369,
      "step": 9906
    },
    {
      "epoch": 0.7141209543718013,
      "grad_norm": 0.12736108780006092,
      "learning_rate": 3.989227808848526e-05,
      "loss": 0.9607,
      "step": 9907
    },
    {
      "epoch": 0.7141930368341383,
      "grad_norm": 0.11763354346714382,
      "learning_rate": 3.987362089495166e-05,
      "loss": 1.006,
      "step": 9908
    },
    {
      "epoch": 0.7142651192964752,
      "grad_norm": 0.2511578665424297,
      "learning_rate": 3.985496697884361e-05,
      "loss": 1.028,
      "step": 9909
    },
    {
      "epoch": 0.7143372017588121,
      "grad_norm": 0.11636358209014039,
      "learning_rate": 3.9836316341177946e-05,
      "loss": 1.0103,
      "step": 9910
    },
    {
      "epoch": 0.714409284221149,
      "grad_norm": 0.1418674862630767,
      "learning_rate": 3.981766898297128e-05,
      "loss": 1.048,
      "step": 9911
    },
    {
      "epoch": 0.7144813666834859,
      "grad_norm": 0.10971505534582053,
      "learning_rate": 3.97990249052401e-05,
      "loss": 1.0591,
      "step": 9912
    },
    {
      "epoch": 0.7145534491458229,
      "grad_norm": 0.17954927617381536,
      "learning_rate": 3.9780384109000556e-05,
      "loss": 1.0502,
      "step": 9913
    },
    {
      "epoch": 0.7146255316081598,
      "grad_norm": 0.16517634248903926,
      "learning_rate": 3.9761746595268865e-05,
      "loss": 0.9844,
      "step": 9914
    },
    {
      "epoch": 0.7146976140704967,
      "grad_norm": 0.12698671619820823,
      "learning_rate": 3.974311236506089e-05,
      "loss": 1.0476,
      "step": 9915
    },
    {
      "epoch": 0.7147696965328335,
      "grad_norm": 0.13652911960651642,
      "learning_rate": 3.972448141939239e-05,
      "loss": 0.969,
      "step": 9916
    },
    {
      "epoch": 0.7148417789951704,
      "grad_norm": 0.10538382656450133,
      "learning_rate": 3.970585375927883e-05,
      "loss": 1.0038,
      "step": 9917
    },
    {
      "epoch": 0.7149138614575073,
      "grad_norm": 0.11436186527065798,
      "learning_rate": 3.968722938573568e-05,
      "loss": 0.9906,
      "step": 9918
    },
    {
      "epoch": 0.7149859439198443,
      "grad_norm": 0.1285471570060189,
      "learning_rate": 3.9668608299778096e-05,
      "loss": 1.0717,
      "step": 9919
    },
    {
      "epoch": 0.7150580263821812,
      "grad_norm": 0.12351883350744661,
      "learning_rate": 3.9649990502421144e-05,
      "loss": 1.0053,
      "step": 9920
    },
    {
      "epoch": 0.7151301088445181,
      "grad_norm": 0.14274985607429685,
      "learning_rate": 3.963137599467958e-05,
      "loss": 1.0222,
      "step": 9921
    },
    {
      "epoch": 0.715202191306855,
      "grad_norm": 0.1411369987735201,
      "learning_rate": 3.9612764777568056e-05,
      "loss": 1.0154,
      "step": 9922
    },
    {
      "epoch": 0.715274273769192,
      "grad_norm": 0.14627153052509362,
      "learning_rate": 3.959415685210114e-05,
      "loss": 1.1013,
      "step": 9923
    },
    {
      "epoch": 0.7153463562315289,
      "grad_norm": 0.10965506743945871,
      "learning_rate": 3.95755522192931e-05,
      "loss": 1.0829,
      "step": 9924
    },
    {
      "epoch": 0.7154184386938658,
      "grad_norm": 0.13381124161277066,
      "learning_rate": 3.9556950880158004e-05,
      "loss": 1.0793,
      "step": 9925
    },
    {
      "epoch": 0.7154905211562027,
      "grad_norm": 0.14621794400994148,
      "learning_rate": 3.95383528357098e-05,
      "loss": 1.1221,
      "step": 9926
    },
    {
      "epoch": 0.7155626036185396,
      "grad_norm": 0.12908129452365988,
      "learning_rate": 3.95197580869623e-05,
      "loss": 0.9581,
      "step": 9927
    },
    {
      "epoch": 0.7156346860808765,
      "grad_norm": 0.134760411255678,
      "learning_rate": 3.950116663492909e-05,
      "loss": 1.0609,
      "step": 9928
    },
    {
      "epoch": 0.7157067685432135,
      "grad_norm": 1.9237025928309874,
      "learning_rate": 3.948257848062351e-05,
      "loss": 1.0216,
      "step": 9929
    },
    {
      "epoch": 0.7157788510055504,
      "grad_norm": 0.11829247579835071,
      "learning_rate": 3.946399362505876e-05,
      "loss": 1.0099,
      "step": 9930
    },
    {
      "epoch": 0.7158509334678873,
      "grad_norm": 0.20226588031612386,
      "learning_rate": 3.9445412069247965e-05,
      "loss": 1.0397,
      "step": 9931
    },
    {
      "epoch": 0.7159230159302242,
      "grad_norm": 0.10845057755870915,
      "learning_rate": 3.942683381420399e-05,
      "loss": 1.049,
      "step": 9932
    },
    {
      "epoch": 0.715995098392561,
      "grad_norm": 0.1161257554602207,
      "learning_rate": 3.940825886093944e-05,
      "loss": 1.0232,
      "step": 9933
    },
    {
      "epoch": 0.716067180854898,
      "grad_norm": 0.12255810869398616,
      "learning_rate": 3.9389687210466855e-05,
      "loss": 1.0045,
      "step": 9934
    },
    {
      "epoch": 0.7161392633172349,
      "grad_norm": 0.12187884436369602,
      "learning_rate": 3.93711188637985e-05,
      "loss": 1.0183,
      "step": 9935
    },
    {
      "epoch": 0.7162113457795718,
      "grad_norm": 0.1152298028438905,
      "learning_rate": 3.935255382194665e-05,
      "loss": 1.1364,
      "step": 9936
    },
    {
      "epoch": 0.7162834282419087,
      "grad_norm": 0.14898380118971485,
      "learning_rate": 3.9333992085923156e-05,
      "loss": 1.0519,
      "step": 9937
    },
    {
      "epoch": 0.7163555107042456,
      "grad_norm": 0.12237813244830995,
      "learning_rate": 3.931543365673982e-05,
      "loss": 0.9909,
      "step": 9938
    },
    {
      "epoch": 0.7164275931665826,
      "grad_norm": 0.11869614033909491,
      "learning_rate": 3.929687853540821e-05,
      "loss": 1.0818,
      "step": 9939
    },
    {
      "epoch": 0.7164996756289195,
      "grad_norm": 0.11369936575718374,
      "learning_rate": 3.927832672293986e-05,
      "loss": 1.0285,
      "step": 9940
    },
    {
      "epoch": 0.7165717580912564,
      "grad_norm": 0.11638619408029842,
      "learning_rate": 3.92597782203459e-05,
      "loss": 1.0181,
      "step": 9941
    },
    {
      "epoch": 0.7166438405535933,
      "grad_norm": 0.13596180829862237,
      "learning_rate": 3.924123302863743e-05,
      "loss": 1.0196,
      "step": 9942
    },
    {
      "epoch": 0.7167159230159302,
      "grad_norm": 0.12398823984958764,
      "learning_rate": 3.922269114882531e-05,
      "loss": 1.0096,
      "step": 9943
    },
    {
      "epoch": 0.7167880054782672,
      "grad_norm": 0.11673709133953444,
      "learning_rate": 3.920415258192026e-05,
      "loss": 1.0827,
      "step": 9944
    },
    {
      "epoch": 0.7168600879406041,
      "grad_norm": 0.12480119362023163,
      "learning_rate": 3.918561732893279e-05,
      "loss": 1.0049,
      "step": 9945
    },
    {
      "epoch": 0.716932170402941,
      "grad_norm": 0.13864365135600876,
      "learning_rate": 3.916708539087324e-05,
      "loss": 1.0308,
      "step": 9946
    },
    {
      "epoch": 0.7170042528652779,
      "grad_norm": 0.12174783122765892,
      "learning_rate": 3.9148556768751754e-05,
      "loss": 1.0194,
      "step": 9947
    },
    {
      "epoch": 0.7170763353276148,
      "grad_norm": 0.12809953430062265,
      "learning_rate": 3.913003146357832e-05,
      "loss": 1.0855,
      "step": 9948
    },
    {
      "epoch": 0.7171484177899518,
      "grad_norm": 0.13330898467912886,
      "learning_rate": 3.911150947636273e-05,
      "loss": 1.0605,
      "step": 9949
    },
    {
      "epoch": 0.7172205002522887,
      "grad_norm": 0.16005694622335495,
      "learning_rate": 3.9092990808114595e-05,
      "loss": 1.0658,
      "step": 9950
    },
    {
      "epoch": 0.7172925827146255,
      "grad_norm": 0.18782710751673945,
      "learning_rate": 3.907447545984335e-05,
      "loss": 1.0631,
      "step": 9951
    },
    {
      "epoch": 0.7173646651769624,
      "grad_norm": 0.10398467936048254,
      "learning_rate": 3.9055963432558265e-05,
      "loss": 1.0204,
      "step": 9952
    },
    {
      "epoch": 0.7174367476392993,
      "grad_norm": 0.13041437544530876,
      "learning_rate": 3.903745472726837e-05,
      "loss": 1.0763,
      "step": 9953
    },
    {
      "epoch": 0.7175088301016362,
      "grad_norm": 0.1463792613569298,
      "learning_rate": 3.901894934498259e-05,
      "loss": 0.9723,
      "step": 9954
    },
    {
      "epoch": 0.7175809125639732,
      "grad_norm": 0.11532957104860724,
      "learning_rate": 3.900044728670963e-05,
      "loss": 1.0143,
      "step": 9955
    },
    {
      "epoch": 0.7176529950263101,
      "grad_norm": 0.12984133558867222,
      "learning_rate": 3.8981948553457994e-05,
      "loss": 1.0529,
      "step": 9956
    },
    {
      "epoch": 0.717725077488647,
      "grad_norm": 0.11050353462234086,
      "learning_rate": 3.896345314623605e-05,
      "loss": 1.0329,
      "step": 9957
    },
    {
      "epoch": 0.7177971599509839,
      "grad_norm": 0.14355493887397203,
      "learning_rate": 3.894496106605197e-05,
      "loss": 1.0976,
      "step": 9958
    },
    {
      "epoch": 0.7178692424133208,
      "grad_norm": 0.2901145667409451,
      "learning_rate": 3.892647231391372e-05,
      "loss": 1.0691,
      "step": 9959
    },
    {
      "epoch": 0.7179413248756578,
      "grad_norm": 0.1381842317930075,
      "learning_rate": 3.890798689082909e-05,
      "loss": 0.9844,
      "step": 9960
    },
    {
      "epoch": 0.7180134073379947,
      "grad_norm": 0.1242375375690978,
      "learning_rate": 3.888950479780574e-05,
      "loss": 1.0137,
      "step": 9961
    },
    {
      "epoch": 0.7180854898003316,
      "grad_norm": 0.1175862792506815,
      "learning_rate": 3.887102603585109e-05,
      "loss": 0.9703,
      "step": 9962
    },
    {
      "epoch": 0.7181575722626685,
      "grad_norm": 0.11930476587356557,
      "learning_rate": 3.8852550605972396e-05,
      "loss": 1.0433,
      "step": 9963
    },
    {
      "epoch": 0.7182296547250054,
      "grad_norm": 0.12200544992136447,
      "learning_rate": 3.883407850917672e-05,
      "loss": 1.0907,
      "step": 9964
    },
    {
      "epoch": 0.7183017371873424,
      "grad_norm": 0.1126979473768788,
      "learning_rate": 3.881560974647101e-05,
      "loss": 1.0413,
      "step": 9965
    },
    {
      "epoch": 0.7183738196496793,
      "grad_norm": 0.12729451435648254,
      "learning_rate": 3.8797144318861875e-05,
      "loss": 1.056,
      "step": 9966
    },
    {
      "epoch": 0.7184459021120162,
      "grad_norm": 0.3041104202822346,
      "learning_rate": 3.8778682227355955e-05,
      "loss": 1.0654,
      "step": 9967
    },
    {
      "epoch": 0.718517984574353,
      "grad_norm": 0.1567722585146074,
      "learning_rate": 3.876022347295955e-05,
      "loss": 1.0718,
      "step": 9968
    },
    {
      "epoch": 0.7185900670366899,
      "grad_norm": 0.11987157123616786,
      "learning_rate": 3.874176805667886e-05,
      "loss": 0.985,
      "step": 9969
    },
    {
      "epoch": 0.7186621494990268,
      "grad_norm": 0.138953654031463,
      "learning_rate": 3.872331597951977e-05,
      "loss": 1.0627,
      "step": 9970
    },
    {
      "epoch": 0.7187342319613638,
      "grad_norm": 0.12313919232030657,
      "learning_rate": 3.87048672424882e-05,
      "loss": 1.0853,
      "step": 9971
    },
    {
      "epoch": 0.7188063144237007,
      "grad_norm": 0.11034710036561143,
      "learning_rate": 3.8686421846589726e-05,
      "loss": 1.0073,
      "step": 9972
    },
    {
      "epoch": 0.7188783968860376,
      "grad_norm": 0.27804729614949686,
      "learning_rate": 3.8667979792829824e-05,
      "loss": 0.9808,
      "step": 9973
    },
    {
      "epoch": 0.7189504793483745,
      "grad_norm": 0.11825078153103863,
      "learning_rate": 3.864954108221367e-05,
      "loss": 0.9974,
      "step": 9974
    },
    {
      "epoch": 0.7190225618107114,
      "grad_norm": 1.0920514386195275,
      "learning_rate": 3.863110571574636e-05,
      "loss": 1.0686,
      "step": 9975
    },
    {
      "epoch": 0.7190946442730484,
      "grad_norm": 0.1268216849478776,
      "learning_rate": 3.861267369443288e-05,
      "loss": 1.0818,
      "step": 9976
    },
    {
      "epoch": 0.7191667267353853,
      "grad_norm": 0.4523397796487868,
      "learning_rate": 3.859424501927783e-05,
      "loss": 1.0056,
      "step": 9977
    },
    {
      "epoch": 0.7192388091977222,
      "grad_norm": 0.11939452943766633,
      "learning_rate": 3.8575819691285783e-05,
      "loss": 1.0268,
      "step": 9978
    },
    {
      "epoch": 0.7193108916600591,
      "grad_norm": 0.12902697015190323,
      "learning_rate": 3.855739771146104e-05,
      "loss": 0.9793,
      "step": 9979
    },
    {
      "epoch": 0.719382974122396,
      "grad_norm": 0.12181981555268559,
      "learning_rate": 3.853897908080788e-05,
      "loss": 0.9806,
      "step": 9980
    },
    {
      "epoch": 0.719455056584733,
      "grad_norm": 0.1217177627324831,
      "learning_rate": 3.8520563800330157e-05,
      "loss": 1.0602,
      "step": 9981
    },
    {
      "epoch": 0.7195271390470699,
      "grad_norm": 0.12356533993571914,
      "learning_rate": 3.850215187103173e-05,
      "loss": 0.9996,
      "step": 9982
    },
    {
      "epoch": 0.7195992215094068,
      "grad_norm": 0.12862321852643838,
      "learning_rate": 3.848374329391617e-05,
      "loss": 1.0902,
      "step": 9983
    },
    {
      "epoch": 0.7196713039717437,
      "grad_norm": 0.11576161412802073,
      "learning_rate": 3.8465338069987e-05,
      "loss": 1.058,
      "step": 9984
    },
    {
      "epoch": 0.7197433864340805,
      "grad_norm": 0.14548700351136964,
      "learning_rate": 3.8446936200247375e-05,
      "loss": 1.106,
      "step": 9985
    },
    {
      "epoch": 0.7198154688964175,
      "grad_norm": 0.12589319173375443,
      "learning_rate": 3.8428537685700406e-05,
      "loss": 0.9656,
      "step": 9986
    },
    {
      "epoch": 0.7198875513587544,
      "grad_norm": 0.11697581276616326,
      "learning_rate": 3.841014252734896e-05,
      "loss": 1.0248,
      "step": 9987
    },
    {
      "epoch": 0.7199596338210913,
      "grad_norm": 0.11146045156691266,
      "learning_rate": 3.839175072619574e-05,
      "loss": 1.0923,
      "step": 9988
    },
    {
      "epoch": 0.7200317162834282,
      "grad_norm": 0.10986673830034614,
      "learning_rate": 3.8373362283243274e-05,
      "loss": 1.0481,
      "step": 9989
    },
    {
      "epoch": 0.7201037987457651,
      "grad_norm": 0.14568531170958313,
      "learning_rate": 3.835497719949389e-05,
      "loss": 1.0702,
      "step": 9990
    },
    {
      "epoch": 0.720175881208102,
      "grad_norm": 0.23424950568295486,
      "learning_rate": 3.8336595475949745e-05,
      "loss": 1.0844,
      "step": 9991
    },
    {
      "epoch": 0.720247963670439,
      "grad_norm": 0.11726107320703713,
      "learning_rate": 3.831821711361281e-05,
      "loss": 0.9819,
      "step": 9992
    },
    {
      "epoch": 0.7203200461327759,
      "grad_norm": 0.1123571847688911,
      "learning_rate": 3.8299842113484855e-05,
      "loss": 1.0414,
      "step": 9993
    },
    {
      "epoch": 0.7203921285951128,
      "grad_norm": 0.15588919911658655,
      "learning_rate": 3.828147047656748e-05,
      "loss": 1.0119,
      "step": 9994
    },
    {
      "epoch": 0.7204642110574497,
      "grad_norm": 0.11916164689791119,
      "learning_rate": 3.826310220386212e-05,
      "loss": 1.0509,
      "step": 9995
    },
    {
      "epoch": 0.7205362935197867,
      "grad_norm": 0.11260929648477563,
      "learning_rate": 3.8244737296370016e-05,
      "loss": 1.0573,
      "step": 9996
    },
    {
      "epoch": 0.7206083759821236,
      "grad_norm": 0.12900813627403418,
      "learning_rate": 3.822637575509219e-05,
      "loss": 1.0443,
      "step": 9997
    },
    {
      "epoch": 0.7206804584444605,
      "grad_norm": 0.20795708190611312,
      "learning_rate": 3.8208017581029545e-05,
      "loss": 1.0318,
      "step": 9998
    },
    {
      "epoch": 0.7207525409067974,
      "grad_norm": 0.12058818715572624,
      "learning_rate": 3.8189662775182734e-05,
      "loss": 1.0295,
      "step": 9999
    },
    {
      "epoch": 0.7208246233691343,
      "grad_norm": 0.17078051602506503,
      "learning_rate": 3.817131133855229e-05,
      "loss": 0.9981,
      "step": 10000
    },
    {
      "epoch": 0.7208967058314713,
      "grad_norm": 0.12141996331418779,
      "learning_rate": 3.81529632721385e-05,
      "loss": 1.0505,
      "step": 10001
    },
    {
      "epoch": 0.7209687882938082,
      "grad_norm": 0.12214491712157001,
      "learning_rate": 3.8134618576941505e-05,
      "loss": 0.9783,
      "step": 10002
    },
    {
      "epoch": 0.721040870756145,
      "grad_norm": 0.17003285244681562,
      "learning_rate": 3.811627725396128e-05,
      "loss": 0.9839,
      "step": 10003
    },
    {
      "epoch": 0.7211129532184819,
      "grad_norm": 0.12026884646652737,
      "learning_rate": 3.809793930419756e-05,
      "loss": 1.0329,
      "step": 10004
    },
    {
      "epoch": 0.7211850356808188,
      "grad_norm": 0.1226173889615984,
      "learning_rate": 3.807960472864994e-05,
      "loss": 1.0067,
      "step": 10005
    },
    {
      "epoch": 0.7212571181431557,
      "grad_norm": 0.11982271163117106,
      "learning_rate": 3.806127352831782e-05,
      "loss": 1.055,
      "step": 10006
    },
    {
      "epoch": 0.7213292006054927,
      "grad_norm": 0.1396158296145009,
      "learning_rate": 3.804294570420041e-05,
      "loss": 1.0233,
      "step": 10007
    },
    {
      "epoch": 0.7214012830678296,
      "grad_norm": 0.11979118778311146,
      "learning_rate": 3.8024621257296747e-05,
      "loss": 1.0725,
      "step": 10008
    },
    {
      "epoch": 0.7214733655301665,
      "grad_norm": 0.1219102714195523,
      "learning_rate": 3.8006300188605705e-05,
      "loss": 1.0956,
      "step": 10009
    },
    {
      "epoch": 0.7215454479925034,
      "grad_norm": 0.15489297816457956,
      "learning_rate": 3.798798249912584e-05,
      "loss": 1.0004,
      "step": 10010
    },
    {
      "epoch": 0.7216175304548403,
      "grad_norm": 0.11518500473961184,
      "learning_rate": 3.796966818985574e-05,
      "loss": 1.0717,
      "step": 10011
    },
    {
      "epoch": 0.7216896129171773,
      "grad_norm": 0.10159903896056773,
      "learning_rate": 3.795135726179367e-05,
      "loss": 0.9971,
      "step": 10012
    },
    {
      "epoch": 0.7217616953795142,
      "grad_norm": 0.14655084002715482,
      "learning_rate": 3.793304971593777e-05,
      "loss": 1.0709,
      "step": 10013
    },
    {
      "epoch": 0.7218337778418511,
      "grad_norm": 0.11573419661168649,
      "learning_rate": 3.7914745553285846e-05,
      "loss": 1.0064,
      "step": 10014
    },
    {
      "epoch": 0.721905860304188,
      "grad_norm": 0.11585498631784558,
      "learning_rate": 3.789644477483576e-05,
      "loss": 0.9705,
      "step": 10015
    },
    {
      "epoch": 0.7219779427665249,
      "grad_norm": 0.12586786917275938,
      "learning_rate": 3.787814738158504e-05,
      "loss": 1.031,
      "step": 10016
    },
    {
      "epoch": 0.7220500252288619,
      "grad_norm": 0.1386508060543989,
      "learning_rate": 3.7859853374531065e-05,
      "loss": 1.1341,
      "step": 10017
    },
    {
      "epoch": 0.7221221076911988,
      "grad_norm": 0.15459775975233941,
      "learning_rate": 3.784156275467097e-05,
      "loss": 1.0845,
      "step": 10018
    },
    {
      "epoch": 0.7221941901535357,
      "grad_norm": 0.14983925055040473,
      "learning_rate": 3.782327552300175e-05,
      "loss": 1.0879,
      "step": 10019
    },
    {
      "epoch": 0.7222662726158725,
      "grad_norm": 0.14759323437322433,
      "learning_rate": 3.780499168052031e-05,
      "loss": 1.0832,
      "step": 10020
    },
    {
      "epoch": 0.7223383550782094,
      "grad_norm": 0.1191640236237747,
      "learning_rate": 3.778671122822327e-05,
      "loss": 1.0168,
      "step": 10021
    },
    {
      "epoch": 0.7224104375405463,
      "grad_norm": 0.1320332825295205,
      "learning_rate": 3.7768434167107006e-05,
      "loss": 1.0153,
      "step": 10022
    },
    {
      "epoch": 0.7224825200028833,
      "grad_norm": 0.1369298141661237,
      "learning_rate": 3.775016049816778e-05,
      "loss": 1.0755,
      "step": 10023
    },
    {
      "epoch": 0.7225546024652202,
      "grad_norm": 0.12149980542498222,
      "learning_rate": 3.7731890222401744e-05,
      "loss": 0.994,
      "step": 10024
    },
    {
      "epoch": 0.7226266849275571,
      "grad_norm": 0.12460634280068444,
      "learning_rate": 3.7713623340804805e-05,
      "loss": 1.0233,
      "step": 10025
    },
    {
      "epoch": 0.722698767389894,
      "grad_norm": 0.10849544707497874,
      "learning_rate": 3.769535985437257e-05,
      "loss": 1.0321,
      "step": 10026
    },
    {
      "epoch": 0.722770849852231,
      "grad_norm": 0.14163896018157873,
      "learning_rate": 3.76770997641006e-05,
      "loss": 0.998,
      "step": 10027
    },
    {
      "epoch": 0.7228429323145679,
      "grad_norm": 0.4102733624583828,
      "learning_rate": 3.765884307098427e-05,
      "loss": 0.9494,
      "step": 10028
    },
    {
      "epoch": 0.7229150147769048,
      "grad_norm": 0.14652927513662103,
      "learning_rate": 3.764058977601875e-05,
      "loss": 0.9648,
      "step": 10029
    },
    {
      "epoch": 0.7229870972392417,
      "grad_norm": 0.1290258873943732,
      "learning_rate": 3.762233988019894e-05,
      "loss": 1.056,
      "step": 10030
    },
    {
      "epoch": 0.7230591797015786,
      "grad_norm": 0.11440708092744076,
      "learning_rate": 3.7604093384519655e-05,
      "loss": 1.023,
      "step": 10031
    },
    {
      "epoch": 0.7231312621639155,
      "grad_norm": 0.12333230201186204,
      "learning_rate": 3.758585028997545e-05,
      "loss": 1.0059,
      "step": 10032
    },
    {
      "epoch": 0.7232033446262525,
      "grad_norm": 0.11961474078396428,
      "learning_rate": 3.7567610597560856e-05,
      "loss": 1.0017,
      "step": 10033
    },
    {
      "epoch": 0.7232754270885894,
      "grad_norm": 0.38270627497963344,
      "learning_rate": 3.754937430826998e-05,
      "loss": 1.012,
      "step": 10034
    },
    {
      "epoch": 0.7233475095509263,
      "grad_norm": 0.13539401173132332,
      "learning_rate": 3.75311414230969e-05,
      "loss": 1.0459,
      "step": 10035
    },
    {
      "epoch": 0.7234195920132632,
      "grad_norm": 0.12354051277395674,
      "learning_rate": 3.7512911943035446e-05,
      "loss": 1.0817,
      "step": 10036
    },
    {
      "epoch": 0.7234916744756,
      "grad_norm": 0.14119566851459459,
      "learning_rate": 3.749468586907937e-05,
      "loss": 1.0364,
      "step": 10037
    },
    {
      "epoch": 0.723563756937937,
      "grad_norm": 0.15110233405775256,
      "learning_rate": 3.747646320222209e-05,
      "loss": 1.1327,
      "step": 10038
    },
    {
      "epoch": 0.7236358394002739,
      "grad_norm": 0.11455944625385099,
      "learning_rate": 3.7458243943456905e-05,
      "loss": 1.0765,
      "step": 10039
    },
    {
      "epoch": 0.7237079218626108,
      "grad_norm": 0.12293649060331305,
      "learning_rate": 3.744002809377693e-05,
      "loss": 1.035,
      "step": 10040
    },
    {
      "epoch": 0.7237800043249477,
      "grad_norm": 0.1625940263306479,
      "learning_rate": 3.742181565417512e-05,
      "loss": 1.0841,
      "step": 10041
    },
    {
      "epoch": 0.7238520867872846,
      "grad_norm": 0.11333355097675861,
      "learning_rate": 3.7403606625644195e-05,
      "loss": 1.0484,
      "step": 10042
    },
    {
      "epoch": 0.7239241692496216,
      "grad_norm": 0.14452398156997076,
      "learning_rate": 3.738540100917671e-05,
      "loss": 1.0291,
      "step": 10043
    },
    {
      "epoch": 0.7239962517119585,
      "grad_norm": 0.12583098077423713,
      "learning_rate": 3.736719880576503e-05,
      "loss": 1.0484,
      "step": 10044
    },
    {
      "epoch": 0.7240683341742954,
      "grad_norm": 0.17409084461033056,
      "learning_rate": 3.734900001640135e-05,
      "loss": 1.0735,
      "step": 10045
    },
    {
      "epoch": 0.7241404166366323,
      "grad_norm": 0.16122138970334945,
      "learning_rate": 3.7330804642077655e-05,
      "loss": 1.0015,
      "step": 10046
    },
    {
      "epoch": 0.7242124990989692,
      "grad_norm": 0.11606208728423317,
      "learning_rate": 3.731261268378576e-05,
      "loss": 1.0806,
      "step": 10047
    },
    {
      "epoch": 0.7242845815613062,
      "grad_norm": 0.12537070563403965,
      "learning_rate": 3.7294424142517304e-05,
      "loss": 1.0482,
      "step": 10048
    },
    {
      "epoch": 0.7243566640236431,
      "grad_norm": 0.12568235475075595,
      "learning_rate": 3.7276239019263705e-05,
      "loss": 0.9848,
      "step": 10049
    },
    {
      "epoch": 0.72442874648598,
      "grad_norm": 0.1153178022945225,
      "learning_rate": 3.7258057315016226e-05,
      "loss": 0.9708,
      "step": 10050
    },
    {
      "epoch": 0.7245008289483169,
      "grad_norm": 0.13864513632726266,
      "learning_rate": 3.723987903076592e-05,
      "loss": 0.9841,
      "step": 10051
    },
    {
      "epoch": 0.7245729114106538,
      "grad_norm": 0.14308115775932087,
      "learning_rate": 3.722170416750368e-05,
      "loss": 0.9763,
      "step": 10052
    },
    {
      "epoch": 0.7246449938729908,
      "grad_norm": 0.1309916768783127,
      "learning_rate": 3.72035327262202e-05,
      "loss": 1.0334,
      "step": 10053
    },
    {
      "epoch": 0.7247170763353277,
      "grad_norm": 0.11626740921789461,
      "learning_rate": 3.718536470790597e-05,
      "loss": 1.0389,
      "step": 10054
    },
    {
      "epoch": 0.7247891587976645,
      "grad_norm": 0.1388517238739733,
      "learning_rate": 3.7167200113551325e-05,
      "loss": 1.0315,
      "step": 10055
    },
    {
      "epoch": 0.7248612412600014,
      "grad_norm": 1.9830092103262522,
      "learning_rate": 3.714903894414638e-05,
      "loss": 1.0317,
      "step": 10056
    },
    {
      "epoch": 0.7249333237223383,
      "grad_norm": 0.1199914636865304,
      "learning_rate": 3.7130881200681115e-05,
      "loss": 1.0809,
      "step": 10057
    },
    {
      "epoch": 0.7250054061846752,
      "grad_norm": 0.11618158125746457,
      "learning_rate": 3.7112726884145254e-05,
      "loss": 0.9452,
      "step": 10058
    },
    {
      "epoch": 0.7250774886470122,
      "grad_norm": 0.12956397410003392,
      "learning_rate": 3.709457599552838e-05,
      "loss": 1.0617,
      "step": 10059
    },
    {
      "epoch": 0.7251495711093491,
      "grad_norm": 0.12674843972999889,
      "learning_rate": 3.7076428535819896e-05,
      "loss": 0.9861,
      "step": 10060
    },
    {
      "epoch": 0.725221653571686,
      "grad_norm": 0.1288270120730414,
      "learning_rate": 3.705828450600898e-05,
      "loss": 1.065,
      "step": 10061
    },
    {
      "epoch": 0.7252937360340229,
      "grad_norm": 0.12154267543386181,
      "learning_rate": 3.704014390708468e-05,
      "loss": 1.022,
      "step": 10062
    },
    {
      "epoch": 0.7253658184963598,
      "grad_norm": 0.11620035337760817,
      "learning_rate": 3.7022006740035734e-05,
      "loss": 1.073,
      "step": 10063
    },
    {
      "epoch": 0.7254379009586968,
      "grad_norm": 0.11352984972206438,
      "learning_rate": 3.7003873005850856e-05,
      "loss": 0.9713,
      "step": 10064
    },
    {
      "epoch": 0.7255099834210337,
      "grad_norm": 0.10345881934883751,
      "learning_rate": 3.698574270551849e-05,
      "loss": 1.0226,
      "step": 10065
    },
    {
      "epoch": 0.7255820658833706,
      "grad_norm": 0.12669226865073271,
      "learning_rate": 3.6967615840026924e-05,
      "loss": 1.0329,
      "step": 10066
    },
    {
      "epoch": 0.7256541483457075,
      "grad_norm": 0.11337259984550706,
      "learning_rate": 3.6949492410364126e-05,
      "loss": 1.0101,
      "step": 10067
    },
    {
      "epoch": 0.7257262308080444,
      "grad_norm": 0.12840145842495035,
      "learning_rate": 3.69313724175181e-05,
      "loss": 1.042,
      "step": 10068
    },
    {
      "epoch": 0.7257983132703814,
      "grad_norm": 0.15325469568892727,
      "learning_rate": 3.6913255862476494e-05,
      "loss": 0.9426,
      "step": 10069
    },
    {
      "epoch": 0.7258703957327183,
      "grad_norm": 0.11330897187644356,
      "learning_rate": 3.6895142746226865e-05,
      "loss": 1.0744,
      "step": 10070
    },
    {
      "epoch": 0.7259424781950552,
      "grad_norm": 0.14109685126571883,
      "learning_rate": 3.687703306975645e-05,
      "loss": 1.0033,
      "step": 10071
    },
    {
      "epoch": 0.726014560657392,
      "grad_norm": 0.10668310909525347,
      "learning_rate": 3.685892683405248e-05,
      "loss": 1.0568,
      "step": 10072
    },
    {
      "epoch": 0.7260866431197289,
      "grad_norm": 0.1244238412037946,
      "learning_rate": 3.684082404010186e-05,
      "loss": 1.1117,
      "step": 10073
    },
    {
      "epoch": 0.7261587255820658,
      "grad_norm": 0.11602737819992255,
      "learning_rate": 3.6822724688891416e-05,
      "loss": 1.0352,
      "step": 10074
    },
    {
      "epoch": 0.7262308080444028,
      "grad_norm": 0.12915715462689295,
      "learning_rate": 3.6804628781407647e-05,
      "loss": 1.0333,
      "step": 10075
    },
    {
      "epoch": 0.7263028905067397,
      "grad_norm": 0.1180729998596275,
      "learning_rate": 3.678653631863693e-05,
      "loss": 1.0267,
      "step": 10076
    },
    {
      "epoch": 0.7263749729690766,
      "grad_norm": 0.12738412890019668,
      "learning_rate": 3.676844730156554e-05,
      "loss": 1.0681,
      "step": 10077
    },
    {
      "epoch": 0.7264470554314135,
      "grad_norm": 0.11840948484332506,
      "learning_rate": 3.6750361731179494e-05,
      "loss": 1.0626,
      "step": 10078
    },
    {
      "epoch": 0.7265191378937504,
      "grad_norm": 0.19617086680184195,
      "learning_rate": 3.673227960846455e-05,
      "loss": 1.042,
      "step": 10079
    },
    {
      "epoch": 0.7265912203560874,
      "grad_norm": 0.11791930714883114,
      "learning_rate": 3.671420093440634e-05,
      "loss": 1.0562,
      "step": 10080
    },
    {
      "epoch": 0.7266633028184243,
      "grad_norm": 0.12033869253580379,
      "learning_rate": 3.669612570999038e-05,
      "loss": 1.0251,
      "step": 10081
    },
    {
      "epoch": 0.7267353852807612,
      "grad_norm": 0.12184838085497075,
      "learning_rate": 3.667805393620194e-05,
      "loss": 1.08,
      "step": 10082
    },
    {
      "epoch": 0.7268074677430981,
      "grad_norm": 0.12473934390197429,
      "learning_rate": 3.665998561402602e-05,
      "loss": 1.1,
      "step": 10083
    },
    {
      "epoch": 0.726879550205435,
      "grad_norm": 0.11441374987803761,
      "learning_rate": 3.6641920744447534e-05,
      "loss": 1.0293,
      "step": 10084
    },
    {
      "epoch": 0.726951632667772,
      "grad_norm": 0.1264217162242873,
      "learning_rate": 3.662385932845116e-05,
      "loss": 1.0444,
      "step": 10085
    },
    {
      "epoch": 0.7270237151301089,
      "grad_norm": 0.12400864482165555,
      "learning_rate": 3.6605801367021494e-05,
      "loss": 1.0084,
      "step": 10086
    },
    {
      "epoch": 0.7270957975924458,
      "grad_norm": 0.11695553555584239,
      "learning_rate": 3.658774686114277e-05,
      "loss": 1.0766,
      "step": 10087
    },
    {
      "epoch": 0.7271678800547827,
      "grad_norm": 0.7233616016260923,
      "learning_rate": 3.656969581179913e-05,
      "loss": 0.9686,
      "step": 10088
    },
    {
      "epoch": 0.7272399625171195,
      "grad_norm": 0.13731414420801424,
      "learning_rate": 3.655164821997451e-05,
      "loss": 0.97,
      "step": 10089
    },
    {
      "epoch": 0.7273120449794565,
      "grad_norm": 0.2370298980304578,
      "learning_rate": 3.653360408665275e-05,
      "loss": 1.0225,
      "step": 10090
    },
    {
      "epoch": 0.7273841274417934,
      "grad_norm": 0.2404755941302694,
      "learning_rate": 3.651556341281732e-05,
      "loss": 1.0397,
      "step": 10091
    },
    {
      "epoch": 0.7274562099041303,
      "grad_norm": 0.15460284974641492,
      "learning_rate": 3.6497526199451626e-05,
      "loss": 0.9848,
      "step": 10092
    },
    {
      "epoch": 0.7275282923664672,
      "grad_norm": 0.11521961057646207,
      "learning_rate": 3.647949244753884e-05,
      "loss": 0.9858,
      "step": 10093
    },
    {
      "epoch": 0.7276003748288041,
      "grad_norm": 0.13491683518569705,
      "learning_rate": 3.6461462158062046e-05,
      "loss": 1.0433,
      "step": 10094
    },
    {
      "epoch": 0.727672457291141,
      "grad_norm": 0.12822140417834704,
      "learning_rate": 3.644343533200397e-05,
      "loss": 1.0498,
      "step": 10095
    },
    {
      "epoch": 0.727744539753478,
      "grad_norm": 0.11822826393587793,
      "learning_rate": 3.642541197034726e-05,
      "loss": 0.9664,
      "step": 10096
    },
    {
      "epoch": 0.7278166222158149,
      "grad_norm": 0.12950508430756322,
      "learning_rate": 3.6407392074074344e-05,
      "loss": 1.0359,
      "step": 10097
    },
    {
      "epoch": 0.7278887046781518,
      "grad_norm": 0.16455986780320925,
      "learning_rate": 3.638937564416748e-05,
      "loss": 1.0582,
      "step": 10098
    },
    {
      "epoch": 0.7279607871404887,
      "grad_norm": 0.12036192339045168,
      "learning_rate": 3.6371362681608725e-05,
      "loss": 1.0396,
      "step": 10099
    },
    {
      "epoch": 0.7280328696028256,
      "grad_norm": 0.12055532359546366,
      "learning_rate": 3.635335318737993e-05,
      "loss": 1.0858,
      "step": 10100
    },
    {
      "epoch": 0.7281049520651626,
      "grad_norm": 0.11054932968942334,
      "learning_rate": 3.633534716246281e-05,
      "loss": 1.0667,
      "step": 10101
    },
    {
      "epoch": 0.7281770345274995,
      "grad_norm": 1.7446872101367046,
      "learning_rate": 3.631734460783881e-05,
      "loss": 1.0297,
      "step": 10102
    },
    {
      "epoch": 0.7282491169898364,
      "grad_norm": 0.11679946704628784,
      "learning_rate": 3.629934552448925e-05,
      "loss": 1.0601,
      "step": 10103
    },
    {
      "epoch": 0.7283211994521733,
      "grad_norm": 0.1397712733532062,
      "learning_rate": 3.628134991339524e-05,
      "loss": 0.9542,
      "step": 10104
    },
    {
      "epoch": 0.7283932819145102,
      "grad_norm": 0.12062643005684397,
      "learning_rate": 3.626335777553772e-05,
      "loss": 1.0489,
      "step": 10105
    },
    {
      "epoch": 0.7284653643768471,
      "grad_norm": 0.13269199696213974,
      "learning_rate": 3.624536911189739e-05,
      "loss": 0.9871,
      "step": 10106
    },
    {
      "epoch": 0.728537446839184,
      "grad_norm": 0.12645685498641251,
      "learning_rate": 3.622738392345482e-05,
      "loss": 1.0415,
      "step": 10107
    },
    {
      "epoch": 0.7286095293015209,
      "grad_norm": 0.10793028190895511,
      "learning_rate": 3.6209402211190337e-05,
      "loss": 1.0129,
      "step": 10108
    },
    {
      "epoch": 0.7286816117638578,
      "grad_norm": 0.12076839201633005,
      "learning_rate": 3.6191423976084126e-05,
      "loss": 1.0661,
      "step": 10109
    },
    {
      "epoch": 0.7287536942261947,
      "grad_norm": 0.15045266011744718,
      "learning_rate": 3.6173449219116164e-05,
      "loss": 1.1957,
      "step": 10110
    },
    {
      "epoch": 0.7288257766885317,
      "grad_norm": 0.13715308735160892,
      "learning_rate": 3.615547794126622e-05,
      "loss": 1.0043,
      "step": 10111
    },
    {
      "epoch": 0.7288978591508686,
      "grad_norm": 0.25708986090052516,
      "learning_rate": 3.6137510143513895e-05,
      "loss": 1.0237,
      "step": 10112
    },
    {
      "epoch": 0.7289699416132055,
      "grad_norm": 0.12115692278946413,
      "learning_rate": 3.6119545826838606e-05,
      "loss": 0.9911,
      "step": 10113
    },
    {
      "epoch": 0.7290420240755424,
      "grad_norm": 0.2897979830724409,
      "learning_rate": 3.610158499221956e-05,
      "loss": 1.0747,
      "step": 10114
    },
    {
      "epoch": 0.7291141065378793,
      "grad_norm": 0.12578821308233792,
      "learning_rate": 3.6083627640635773e-05,
      "loss": 1.0171,
      "step": 10115
    },
    {
      "epoch": 0.7291861890002163,
      "grad_norm": 0.108486198720806,
      "learning_rate": 3.606567377306611e-05,
      "loss": 1.0681,
      "step": 10116
    },
    {
      "epoch": 0.7292582714625532,
      "grad_norm": 0.12612071653624624,
      "learning_rate": 3.604772339048919e-05,
      "loss": 1.0073,
      "step": 10117
    },
    {
      "epoch": 0.7293303539248901,
      "grad_norm": 0.9602515987538939,
      "learning_rate": 3.602977649388347e-05,
      "loss": 1.0783,
      "step": 10118
    },
    {
      "epoch": 0.729402436387227,
      "grad_norm": 0.12790985388996626,
      "learning_rate": 3.6011833084227266e-05,
      "loss": 1.0182,
      "step": 10119
    },
    {
      "epoch": 0.7294745188495639,
      "grad_norm": 0.1188787962121318,
      "learning_rate": 3.599389316249855e-05,
      "loss": 1.0214,
      "step": 10120
    },
    {
      "epoch": 0.7295466013119009,
      "grad_norm": 0.13455380669529127,
      "learning_rate": 3.5975956729675294e-05,
      "loss": 1.0319,
      "step": 10121
    },
    {
      "epoch": 0.7296186837742378,
      "grad_norm": 0.14609386869426164,
      "learning_rate": 3.5958023786735186e-05,
      "loss": 0.9972,
      "step": 10122
    },
    {
      "epoch": 0.7296907662365747,
      "grad_norm": 0.11864548148795685,
      "learning_rate": 3.594009433465575e-05,
      "loss": 1.1149,
      "step": 10123
    },
    {
      "epoch": 0.7297628486989115,
      "grad_norm": 0.14710191926335725,
      "learning_rate": 3.592216837441419e-05,
      "loss": 0.9875,
      "step": 10124
    },
    {
      "epoch": 0.7298349311612484,
      "grad_norm": 0.2186694266031524,
      "learning_rate": 3.590424590698775e-05,
      "loss": 1.046,
      "step": 10125
    },
    {
      "epoch": 0.7299070136235853,
      "grad_norm": 0.147246939998329,
      "learning_rate": 3.588632693335332e-05,
      "loss": 1.0747,
      "step": 10126
    },
    {
      "epoch": 0.7299790960859223,
      "grad_norm": 0.15364282482835412,
      "learning_rate": 3.586841145448769e-05,
      "loss": 1.1364,
      "step": 10127
    },
    {
      "epoch": 0.7300511785482592,
      "grad_norm": 0.14279287830648305,
      "learning_rate": 3.585049947136734e-05,
      "loss": 1.0724,
      "step": 10128
    },
    {
      "epoch": 0.7301232610105961,
      "grad_norm": 0.11328021310351684,
      "learning_rate": 3.583259098496863e-05,
      "loss": 1.0505,
      "step": 10129
    },
    {
      "epoch": 0.730195343472933,
      "grad_norm": 0.15353958674748472,
      "learning_rate": 3.581468599626784e-05,
      "loss": 1.0225,
      "step": 10130
    },
    {
      "epoch": 0.7302674259352699,
      "grad_norm": 0.13830697822431123,
      "learning_rate": 3.579678450624084e-05,
      "loss": 0.9936,
      "step": 10131
    },
    {
      "epoch": 0.7303395083976069,
      "grad_norm": 0.14305422820636537,
      "learning_rate": 3.5778886515863474e-05,
      "loss": 1.0547,
      "step": 10132
    },
    {
      "epoch": 0.7304115908599438,
      "grad_norm": 0.11599997438817816,
      "learning_rate": 3.5760992026111285e-05,
      "loss": 1.0708,
      "step": 10133
    },
    {
      "epoch": 0.7304836733222807,
      "grad_norm": 0.10456339545229815,
      "learning_rate": 3.5743101037959805e-05,
      "loss": 1.0762,
      "step": 10134
    },
    {
      "epoch": 0.7305557557846176,
      "grad_norm": 0.1267531558081211,
      "learning_rate": 3.572521355238414e-05,
      "loss": 1.0292,
      "step": 10135
    },
    {
      "epoch": 0.7306278382469545,
      "grad_norm": 0.12237888696843394,
      "learning_rate": 3.570732957035936e-05,
      "loss": 1.0749,
      "step": 10136
    },
    {
      "epoch": 0.7306999207092915,
      "grad_norm": 0.14403095706819408,
      "learning_rate": 3.568944909286026e-05,
      "loss": 1.0532,
      "step": 10137
    },
    {
      "epoch": 0.7307720031716284,
      "grad_norm": 0.1095934177948193,
      "learning_rate": 3.5671572120861596e-05,
      "loss": 1.1488,
      "step": 10138
    },
    {
      "epoch": 0.7308440856339653,
      "grad_norm": 0.12805380692154616,
      "learning_rate": 3.565369865533771e-05,
      "loss": 1.0812,
      "step": 10139
    },
    {
      "epoch": 0.7309161680963022,
      "grad_norm": 0.15695243630876923,
      "learning_rate": 3.5635828697262904e-05,
      "loss": 1.0191,
      "step": 10140
    },
    {
      "epoch": 0.730988250558639,
      "grad_norm": 0.1170518061012586,
      "learning_rate": 3.561796224761126e-05,
      "loss": 1.0157,
      "step": 10141
    },
    {
      "epoch": 0.731060333020976,
      "grad_norm": 0.1233640213671953,
      "learning_rate": 3.560009930735664e-05,
      "loss": 1.063,
      "step": 10142
    },
    {
      "epoch": 0.7311324154833129,
      "grad_norm": 0.11677840046200036,
      "learning_rate": 3.5582239877472756e-05,
      "loss": 1.0339,
      "step": 10143
    },
    {
      "epoch": 0.7312044979456498,
      "grad_norm": 0.11704470693428087,
      "learning_rate": 3.556438395893309e-05,
      "loss": 1.047,
      "step": 10144
    },
    {
      "epoch": 0.7312765804079867,
      "grad_norm": 0.12314066655997187,
      "learning_rate": 3.554653155271096e-05,
      "loss": 0.968,
      "step": 10145
    },
    {
      "epoch": 0.7313486628703236,
      "grad_norm": 0.12266229740900464,
      "learning_rate": 3.552868265977948e-05,
      "loss": 0.9735,
      "step": 10146
    },
    {
      "epoch": 0.7314207453326605,
      "grad_norm": 0.36302006485493676,
      "learning_rate": 3.551083728111157e-05,
      "loss": 1.0081,
      "step": 10147
    },
    {
      "epoch": 0.7314928277949975,
      "grad_norm": 0.12985280025053017,
      "learning_rate": 3.5492995417679956e-05,
      "loss": 1.0422,
      "step": 10148
    },
    {
      "epoch": 0.7315649102573344,
      "grad_norm": 0.12405116863168006,
      "learning_rate": 3.547515707045719e-05,
      "loss": 1.036,
      "step": 10149
    },
    {
      "epoch": 0.7316369927196713,
      "grad_norm": 0.1277426367285499,
      "learning_rate": 3.545732224041562e-05,
      "loss": 1.0899,
      "step": 10150
    },
    {
      "epoch": 0.7317090751820082,
      "grad_norm": 0.12524976298621068,
      "learning_rate": 3.543949092852741e-05,
      "loss": 1.0616,
      "step": 10151
    },
    {
      "epoch": 0.7317811576443451,
      "grad_norm": 0.11560345811524955,
      "learning_rate": 3.542166313576451e-05,
      "loss": 1.0969,
      "step": 10152
    },
    {
      "epoch": 0.7318532401066821,
      "grad_norm": 0.1316770041976106,
      "learning_rate": 3.5403838863098715e-05,
      "loss": 0.9672,
      "step": 10153
    },
    {
      "epoch": 0.731925322569019,
      "grad_norm": 0.14435286468367878,
      "learning_rate": 3.538601811150159e-05,
      "loss": 1.0473,
      "step": 10154
    },
    {
      "epoch": 0.7319974050313559,
      "grad_norm": 0.11288241086029892,
      "learning_rate": 3.536820088194453e-05,
      "loss": 1.1079,
      "step": 10155
    },
    {
      "epoch": 0.7320694874936928,
      "grad_norm": 0.13130649012440673,
      "learning_rate": 3.535038717539875e-05,
      "loss": 1.047,
      "step": 10156
    },
    {
      "epoch": 0.7321415699560297,
      "grad_norm": 0.11700777231744482,
      "learning_rate": 3.533257699283522e-05,
      "loss": 0.9707,
      "step": 10157
    },
    {
      "epoch": 0.7322136524183666,
      "grad_norm": 0.1643219735760751,
      "learning_rate": 3.5314770335224775e-05,
      "loss": 1.0732,
      "step": 10158
    },
    {
      "epoch": 0.7322857348807035,
      "grad_norm": 0.11413755856136053,
      "learning_rate": 3.529696720353805e-05,
      "loss": 1.0335,
      "step": 10159
    },
    {
      "epoch": 0.7323578173430404,
      "grad_norm": 0.12038253771031354,
      "learning_rate": 3.5279167598745455e-05,
      "loss": 0.9694,
      "step": 10160
    },
    {
      "epoch": 0.7324298998053773,
      "grad_norm": 0.23128867260705901,
      "learning_rate": 3.5261371521817244e-05,
      "loss": 1.0418,
      "step": 10161
    },
    {
      "epoch": 0.7325019822677142,
      "grad_norm": 0.1086054125021735,
      "learning_rate": 3.5243578973723444e-05,
      "loss": 1.1046,
      "step": 10162
    },
    {
      "epoch": 0.7325740647300512,
      "grad_norm": 0.389484424220679,
      "learning_rate": 3.5225789955433954e-05,
      "loss": 1.112,
      "step": 10163
    },
    {
      "epoch": 0.7326461471923881,
      "grad_norm": 0.11221401666822002,
      "learning_rate": 3.5208004467918324e-05,
      "loss": 1.0364,
      "step": 10164
    },
    {
      "epoch": 0.732718229654725,
      "grad_norm": 0.11668283610211617,
      "learning_rate": 3.519022251214614e-05,
      "loss": 1.0334,
      "step": 10165
    },
    {
      "epoch": 0.7327903121170619,
      "grad_norm": 0.13777828526029504,
      "learning_rate": 3.517244408908663e-05,
      "loss": 0.9881,
      "step": 10166
    },
    {
      "epoch": 0.7328623945793988,
      "grad_norm": 0.18625622544588766,
      "learning_rate": 3.5154669199708915e-05,
      "loss": 0.9869,
      "step": 10167
    },
    {
      "epoch": 0.7329344770417358,
      "grad_norm": 0.12679020635120636,
      "learning_rate": 3.513689784498178e-05,
      "loss": 1.064,
      "step": 10168
    },
    {
      "epoch": 0.7330065595040727,
      "grad_norm": 0.165839448317683,
      "learning_rate": 3.5119130025874034e-05,
      "loss": 1.0513,
      "step": 10169
    },
    {
      "epoch": 0.7330786419664096,
      "grad_norm": 0.1556930779598024,
      "learning_rate": 3.510136574335414e-05,
      "loss": 1.0877,
      "step": 10170
    },
    {
      "epoch": 0.7331507244287465,
      "grad_norm": 0.13353874228569884,
      "learning_rate": 3.508360499839044e-05,
      "loss": 1.1442,
      "step": 10171
    },
    {
      "epoch": 0.7332228068910834,
      "grad_norm": 0.11410663230864708,
      "learning_rate": 3.5065847791951e-05,
      "loss": 1.0622,
      "step": 10172
    },
    {
      "epoch": 0.7332948893534204,
      "grad_norm": 0.5278140568155396,
      "learning_rate": 3.504809412500374e-05,
      "loss": 1.0181,
      "step": 10173
    },
    {
      "epoch": 0.7333669718157573,
      "grad_norm": 0.11837233210369108,
      "learning_rate": 3.5030343998516455e-05,
      "loss": 0.961,
      "step": 10174
    },
    {
      "epoch": 0.7334390542780942,
      "grad_norm": 0.11541499683596897,
      "learning_rate": 3.5012597413456706e-05,
      "loss": 0.9544,
      "step": 10175
    },
    {
      "epoch": 0.733511136740431,
      "grad_norm": 0.1268157731713385,
      "learning_rate": 3.499485437079174e-05,
      "loss": 1.0139,
      "step": 10176
    },
    {
      "epoch": 0.7335832192027679,
      "grad_norm": 0.115846108933764,
      "learning_rate": 3.497711487148874e-05,
      "loss": 1.0439,
      "step": 10177
    },
    {
      "epoch": 0.7336553016651048,
      "grad_norm": 0.10252309154642932,
      "learning_rate": 3.495937891651472e-05,
      "loss": 1.0748,
      "step": 10178
    },
    {
      "epoch": 0.7337273841274418,
      "grad_norm": 0.11979161466073707,
      "learning_rate": 3.494164650683646e-05,
      "loss": 1.0027,
      "step": 10179
    },
    {
      "epoch": 0.7337994665897787,
      "grad_norm": 0.11280229859366264,
      "learning_rate": 3.492391764342044e-05,
      "loss": 1.0252,
      "step": 10180
    },
    {
      "epoch": 0.7338715490521156,
      "grad_norm": 0.15766621456668145,
      "learning_rate": 3.490619232723308e-05,
      "loss": 1.0711,
      "step": 10181
    },
    {
      "epoch": 0.7339436315144525,
      "grad_norm": 0.1373772264660704,
      "learning_rate": 3.4888470559240616e-05,
      "loss": 1.0948,
      "step": 10182
    },
    {
      "epoch": 0.7340157139767894,
      "grad_norm": 0.12629049280307586,
      "learning_rate": 3.4870752340409043e-05,
      "loss": 0.9959,
      "step": 10183
    },
    {
      "epoch": 0.7340877964391264,
      "grad_norm": 0.13664234281835258,
      "learning_rate": 3.4853037671704094e-05,
      "loss": 0.9685,
      "step": 10184
    },
    {
      "epoch": 0.7341598789014633,
      "grad_norm": 0.14635045352672624,
      "learning_rate": 3.483532655409142e-05,
      "loss": 0.9983,
      "step": 10185
    },
    {
      "epoch": 0.7342319613638002,
      "grad_norm": 0.13433262884366928,
      "learning_rate": 3.4817618988536394e-05,
      "loss": 1.0548,
      "step": 10186
    },
    {
      "epoch": 0.7343040438261371,
      "grad_norm": 0.12723341390030438,
      "learning_rate": 3.4799914976004336e-05,
      "loss": 1.0262,
      "step": 10187
    },
    {
      "epoch": 0.734376126288474,
      "grad_norm": 0.13157746407684096,
      "learning_rate": 3.4782214517460165e-05,
      "loss": 0.9896,
      "step": 10188
    },
    {
      "epoch": 0.734448208750811,
      "grad_norm": 0.19009458555762004,
      "learning_rate": 3.476451761386876e-05,
      "loss": 1.0291,
      "step": 10189
    },
    {
      "epoch": 0.7345202912131479,
      "grad_norm": 0.12387861002381799,
      "learning_rate": 3.4746824266194744e-05,
      "loss": 1.0077,
      "step": 10190
    },
    {
      "epoch": 0.7345923736754848,
      "grad_norm": 0.11312128301206865,
      "learning_rate": 3.472913447540263e-05,
      "loss": 0.9817,
      "step": 10191
    },
    {
      "epoch": 0.7346644561378217,
      "grad_norm": 0.11904362328060437,
      "learning_rate": 3.471144824245658e-05,
      "loss": 0.961,
      "step": 10192
    },
    {
      "epoch": 0.7347365386001585,
      "grad_norm": 0.11651149115541226,
      "learning_rate": 3.4693765568320694e-05,
      "loss": 1.0206,
      "step": 10193
    },
    {
      "epoch": 0.7348086210624954,
      "grad_norm": 0.11506755670312237,
      "learning_rate": 3.467608645395883e-05,
      "loss": 1.0353,
      "step": 10194
    },
    {
      "epoch": 0.7348807035248324,
      "grad_norm": 0.14476540398928223,
      "learning_rate": 3.465841090033466e-05,
      "loss": 1.032,
      "step": 10195
    },
    {
      "epoch": 0.7349527859871693,
      "grad_norm": 0.12608942161555545,
      "learning_rate": 3.4640738908411664e-05,
      "loss": 1.0558,
      "step": 10196
    },
    {
      "epoch": 0.7350248684495062,
      "grad_norm": 0.11905950916829065,
      "learning_rate": 3.462307047915311e-05,
      "loss": 1.0576,
      "step": 10197
    },
    {
      "epoch": 0.7350969509118431,
      "grad_norm": 0.1727465108864931,
      "learning_rate": 3.460540561352209e-05,
      "loss": 1.013,
      "step": 10198
    },
    {
      "epoch": 0.73516903337418,
      "grad_norm": 0.13307198207106588,
      "learning_rate": 3.458774431248151e-05,
      "loss": 1.0913,
      "step": 10199
    },
    {
      "epoch": 0.735241115836517,
      "grad_norm": 0.11063854648525473,
      "learning_rate": 3.457008657699405e-05,
      "loss": 1.0459,
      "step": 10200
    },
    {
      "epoch": 0.7353131982988539,
      "grad_norm": 0.16823671630888057,
      "learning_rate": 3.455243240802224e-05,
      "loss": 1.0663,
      "step": 10201
    },
    {
      "epoch": 0.7353852807611908,
      "grad_norm": 0.1436018362718538,
      "learning_rate": 3.453478180652836e-05,
      "loss": 1.004,
      "step": 10202
    },
    {
      "epoch": 0.7354573632235277,
      "grad_norm": 0.10410471853241506,
      "learning_rate": 3.451713477347455e-05,
      "loss": 1.1289,
      "step": 10203
    },
    {
      "epoch": 0.7355294456858646,
      "grad_norm": 0.10780961432480607,
      "learning_rate": 3.449949130982271e-05,
      "loss": 1.0588,
      "step": 10204
    },
    {
      "epoch": 0.7356015281482016,
      "grad_norm": 0.1227800726690441,
      "learning_rate": 3.448185141653458e-05,
      "loss": 1.086,
      "step": 10205
    },
    {
      "epoch": 0.7356736106105385,
      "grad_norm": 0.11658547853693896,
      "learning_rate": 3.446421509457169e-05,
      "loss": 1.0897,
      "step": 10206
    },
    {
      "epoch": 0.7357456930728754,
      "grad_norm": 0.13418295138542366,
      "learning_rate": 3.44465823448954e-05,
      "loss": 0.9904,
      "step": 10207
    },
    {
      "epoch": 0.7358177755352123,
      "grad_norm": 0.11281591094424118,
      "learning_rate": 3.4428953168466806e-05,
      "loss": 0.9918,
      "step": 10208
    },
    {
      "epoch": 0.7358898579975492,
      "grad_norm": 0.13343136403567402,
      "learning_rate": 3.441132756624687e-05,
      "loss": 1.0355,
      "step": 10209
    },
    {
      "epoch": 0.735961940459886,
      "grad_norm": 0.1366153497108959,
      "learning_rate": 3.4393705539196376e-05,
      "loss": 1.0144,
      "step": 10210
    },
    {
      "epoch": 0.736034022922223,
      "grad_norm": 0.1211582899233603,
      "learning_rate": 3.437608708827585e-05,
      "loss": 1.0896,
      "step": 10211
    },
    {
      "epoch": 0.7361061053845599,
      "grad_norm": 0.11594050310193868,
      "learning_rate": 3.435847221444568e-05,
      "loss": 1.0324,
      "step": 10212
    },
    {
      "epoch": 0.7361781878468968,
      "grad_norm": 0.18858887729436816,
      "learning_rate": 3.434086091866601e-05,
      "loss": 1.0965,
      "step": 10213
    },
    {
      "epoch": 0.7362502703092337,
      "grad_norm": 0.12415127427229038,
      "learning_rate": 3.4323253201896824e-05,
      "loss": 1.01,
      "step": 10214
    },
    {
      "epoch": 0.7363223527715707,
      "grad_norm": 0.1302045772433785,
      "learning_rate": 3.43056490650979e-05,
      "loss": 1.1062,
      "step": 10215
    },
    {
      "epoch": 0.7363944352339076,
      "grad_norm": 0.15912917853167363,
      "learning_rate": 3.4288048509228855e-05,
      "loss": 1.0797,
      "step": 10216
    },
    {
      "epoch": 0.7364665176962445,
      "grad_norm": 0.14639358836606936,
      "learning_rate": 3.427045153524897e-05,
      "loss": 1.065,
      "step": 10217
    },
    {
      "epoch": 0.7365386001585814,
      "grad_norm": 0.10566456162912784,
      "learning_rate": 3.425285814411755e-05,
      "loss": 1.058,
      "step": 10218
    },
    {
      "epoch": 0.7366106826209183,
      "grad_norm": 0.6377174293396557,
      "learning_rate": 3.423526833679355e-05,
      "loss": 1.12,
      "step": 10219
    },
    {
      "epoch": 0.7366827650832553,
      "grad_norm": 0.2210944754708561,
      "learning_rate": 3.42176821142358e-05,
      "loss": 0.9973,
      "step": 10220
    },
    {
      "epoch": 0.7367548475455922,
      "grad_norm": 0.15634231369505255,
      "learning_rate": 3.4200099477402814e-05,
      "loss": 0.9801,
      "step": 10221
    },
    {
      "epoch": 0.7368269300079291,
      "grad_norm": 0.3288896680415189,
      "learning_rate": 3.418252042725312e-05,
      "loss": 0.982,
      "step": 10222
    },
    {
      "epoch": 0.736899012470266,
      "grad_norm": 0.11666779144017522,
      "learning_rate": 3.416494496474486e-05,
      "loss": 1.0621,
      "step": 10223
    },
    {
      "epoch": 0.7369710949326029,
      "grad_norm": 0.12433748295466031,
      "learning_rate": 3.414737309083612e-05,
      "loss": 0.9726,
      "step": 10224
    },
    {
      "epoch": 0.7370431773949399,
      "grad_norm": 0.12060810407403273,
      "learning_rate": 3.4129804806484624e-05,
      "loss": 1.0728,
      "step": 10225
    },
    {
      "epoch": 0.7371152598572768,
      "grad_norm": 0.12265375892559485,
      "learning_rate": 3.4112240112648075e-05,
      "loss": 1.0395,
      "step": 10226
    },
    {
      "epoch": 0.7371873423196137,
      "grad_norm": 0.3098689398852182,
      "learning_rate": 3.4094679010283895e-05,
      "loss": 1.0854,
      "step": 10227
    },
    {
      "epoch": 0.7372594247819505,
      "grad_norm": 0.1322648828167549,
      "learning_rate": 3.407712150034935e-05,
      "loss": 1.0461,
      "step": 10228
    },
    {
      "epoch": 0.7373315072442874,
      "grad_norm": 0.11023855232318823,
      "learning_rate": 3.405956758380141e-05,
      "loss": 0.9732,
      "step": 10229
    },
    {
      "epoch": 0.7374035897066243,
      "grad_norm": 0.12531849558572242,
      "learning_rate": 3.404201726159693e-05,
      "loss": 0.9508,
      "step": 10230
    },
    {
      "epoch": 0.7374756721689613,
      "grad_norm": 0.19696403606385718,
      "learning_rate": 3.402447053469261e-05,
      "loss": 1.0522,
      "step": 10231
    },
    {
      "epoch": 0.7375477546312982,
      "grad_norm": 0.14313499895200507,
      "learning_rate": 3.4006927404044916e-05,
      "loss": 1.0371,
      "step": 10232
    },
    {
      "epoch": 0.7376198370936351,
      "grad_norm": 0.1147793573888027,
      "learning_rate": 3.3989387870610036e-05,
      "loss": 1.0514,
      "step": 10233
    },
    {
      "epoch": 0.737691919555972,
      "grad_norm": 0.1366384318692159,
      "learning_rate": 3.397185193534403e-05,
      "loss": 1.0261,
      "step": 10234
    },
    {
      "epoch": 0.7377640020183089,
      "grad_norm": 0.1307443702399348,
      "learning_rate": 3.395431959920283e-05,
      "loss": 1.0219,
      "step": 10235
    },
    {
      "epoch": 0.7378360844806459,
      "grad_norm": 1.0989236854413378,
      "learning_rate": 3.393679086314212e-05,
      "loss": 1.0451,
      "step": 10236
    },
    {
      "epoch": 0.7379081669429828,
      "grad_norm": 0.12501846901803138,
      "learning_rate": 3.391926572811728e-05,
      "loss": 1.0549,
      "step": 10237
    },
    {
      "epoch": 0.7379802494053197,
      "grad_norm": 0.1425165696803415,
      "learning_rate": 3.390174419508364e-05,
      "loss": 1.1003,
      "step": 10238
    },
    {
      "epoch": 0.7380523318676566,
      "grad_norm": 0.13510389382969365,
      "learning_rate": 3.388422626499623e-05,
      "loss": 1.0586,
      "step": 10239
    },
    {
      "epoch": 0.7381244143299935,
      "grad_norm": 0.1486095710769688,
      "learning_rate": 3.386671193881006e-05,
      "loss": 1.055,
      "step": 10240
    },
    {
      "epoch": 0.7381964967923305,
      "grad_norm": 0.19185183608153708,
      "learning_rate": 3.3849201217479684e-05,
      "loss": 1.0954,
      "step": 10241
    },
    {
      "epoch": 0.7382685792546674,
      "grad_norm": 0.11773883041494192,
      "learning_rate": 3.383169410195965e-05,
      "loss": 0.9643,
      "step": 10242
    },
    {
      "epoch": 0.7383406617170043,
      "grad_norm": 0.11251591659267085,
      "learning_rate": 3.381419059320422e-05,
      "loss": 1.0658,
      "step": 10243
    },
    {
      "epoch": 0.7384127441793412,
      "grad_norm": 0.12339549875151982,
      "learning_rate": 3.3796690692167576e-05,
      "loss": 1.0605,
      "step": 10244
    },
    {
      "epoch": 0.738484826641678,
      "grad_norm": 0.12462273904333211,
      "learning_rate": 3.3779194399803536e-05,
      "loss": 1.1006,
      "step": 10245
    },
    {
      "epoch": 0.738556909104015,
      "grad_norm": 0.12074536900937387,
      "learning_rate": 3.376170171706582e-05,
      "loss": 0.9728,
      "step": 10246
    },
    {
      "epoch": 0.7386289915663519,
      "grad_norm": 0.1248754598069677,
      "learning_rate": 3.374421264490795e-05,
      "loss": 1.0312,
      "step": 10247
    },
    {
      "epoch": 0.7387010740286888,
      "grad_norm": 0.13619609120469975,
      "learning_rate": 3.3726727184283236e-05,
      "loss": 1.0519,
      "step": 10248
    },
    {
      "epoch": 0.7387731564910257,
      "grad_norm": 0.11266374068803682,
      "learning_rate": 3.3709245336144777e-05,
      "loss": 0.9971,
      "step": 10249
    },
    {
      "epoch": 0.7388452389533626,
      "grad_norm": 0.1509252654010762,
      "learning_rate": 3.36917671014455e-05,
      "loss": 1.0359,
      "step": 10250
    },
    {
      "epoch": 0.7389173214156995,
      "grad_norm": 0.1357036780116914,
      "learning_rate": 3.367429248113815e-05,
      "loss": 1.0542,
      "step": 10251
    },
    {
      "epoch": 0.7389894038780365,
      "grad_norm": 0.11912064242389404,
      "learning_rate": 3.365682147617521e-05,
      "loss": 1.0745,
      "step": 10252
    },
    {
      "epoch": 0.7390614863403734,
      "grad_norm": 0.11796458720549045,
      "learning_rate": 3.363935408750901e-05,
      "loss": 1.0188,
      "step": 10253
    },
    {
      "epoch": 0.7391335688027103,
      "grad_norm": 0.16756419358415303,
      "learning_rate": 3.3621890316091706e-05,
      "loss": 0.999,
      "step": 10254
    },
    {
      "epoch": 0.7392056512650472,
      "grad_norm": 0.1407912929886326,
      "learning_rate": 3.360443016287522e-05,
      "loss": 1.005,
      "step": 10255
    },
    {
      "epoch": 0.7392777337273841,
      "grad_norm": 0.13398348224113546,
      "learning_rate": 3.358697362881127e-05,
      "loss": 1.0155,
      "step": 10256
    },
    {
      "epoch": 0.7393498161897211,
      "grad_norm": 0.13553451052765775,
      "learning_rate": 3.356952071485141e-05,
      "loss": 1.0301,
      "step": 10257
    },
    {
      "epoch": 0.739421898652058,
      "grad_norm": 0.14387582939189725,
      "learning_rate": 3.3552071421946983e-05,
      "loss": 1.001,
      "step": 10258
    },
    {
      "epoch": 0.7394939811143949,
      "grad_norm": 0.11625440880948339,
      "learning_rate": 3.353462575104911e-05,
      "loss": 1.0136,
      "step": 10259
    },
    {
      "epoch": 0.7395660635767318,
      "grad_norm": 0.12264834017697877,
      "learning_rate": 3.351718370310876e-05,
      "loss": 1.0224,
      "step": 10260
    },
    {
      "epoch": 0.7396381460390687,
      "grad_norm": 0.1066914338659276,
      "learning_rate": 3.3499745279076665e-05,
      "loss": 1.0477,
      "step": 10261
    },
    {
      "epoch": 0.7397102285014056,
      "grad_norm": 0.1305944527949246,
      "learning_rate": 3.3482310479903386e-05,
      "loss": 0.9834,
      "step": 10262
    },
    {
      "epoch": 0.7397823109637425,
      "grad_norm": 0.12195424046497023,
      "learning_rate": 3.3464879306539267e-05,
      "loss": 1.0759,
      "step": 10263
    },
    {
      "epoch": 0.7398543934260794,
      "grad_norm": 0.11308462441030859,
      "learning_rate": 3.344745175993447e-05,
      "loss": 0.9313,
      "step": 10264
    },
    {
      "epoch": 0.7399264758884163,
      "grad_norm": 0.14410280586832205,
      "learning_rate": 3.343002784103893e-05,
      "loss": 1.0777,
      "step": 10265
    },
    {
      "epoch": 0.7399985583507532,
      "grad_norm": 0.16510991406131253,
      "learning_rate": 3.3412607550802444e-05,
      "loss": 1.0872,
      "step": 10266
    },
    {
      "epoch": 0.7400706408130902,
      "grad_norm": 0.14198437706909728,
      "learning_rate": 3.339519089017455e-05,
      "loss": 1.0393,
      "step": 10267
    },
    {
      "epoch": 0.7401427232754271,
      "grad_norm": 0.12103056111536968,
      "learning_rate": 3.337777786010461e-05,
      "loss": 1.0029,
      "step": 10268
    },
    {
      "epoch": 0.740214805737764,
      "grad_norm": 0.12317898545970216,
      "learning_rate": 3.3360368461541835e-05,
      "loss": 1.0249,
      "step": 10269
    },
    {
      "epoch": 0.7402868882001009,
      "grad_norm": 0.22512212526082218,
      "learning_rate": 3.334296269543509e-05,
      "loss": 0.959,
      "step": 10270
    },
    {
      "epoch": 0.7403589706624378,
      "grad_norm": 0.1274626659836836,
      "learning_rate": 3.332556056273324e-05,
      "loss": 1.0598,
      "step": 10271
    },
    {
      "epoch": 0.7404310531247748,
      "grad_norm": 0.19481424106348627,
      "learning_rate": 3.330816206438482e-05,
      "loss": 1.0642,
      "step": 10272
    },
    {
      "epoch": 0.7405031355871117,
      "grad_norm": 0.4827219328290077,
      "learning_rate": 3.329076720133825e-05,
      "loss": 0.9709,
      "step": 10273
    },
    {
      "epoch": 0.7405752180494486,
      "grad_norm": 0.13638897491285787,
      "learning_rate": 3.32733759745416e-05,
      "loss": 1.0157,
      "step": 10274
    },
    {
      "epoch": 0.7406473005117855,
      "grad_norm": 0.1281584397093393,
      "learning_rate": 3.325598838494294e-05,
      "loss": 1.0891,
      "step": 10275
    },
    {
      "epoch": 0.7407193829741224,
      "grad_norm": 0.15806220724164363,
      "learning_rate": 3.323860443349003e-05,
      "loss": 1.1176,
      "step": 10276
    },
    {
      "epoch": 0.7407914654364594,
      "grad_norm": 0.2067620486668828,
      "learning_rate": 3.322122412113047e-05,
      "loss": 1.0946,
      "step": 10277
    },
    {
      "epoch": 0.7408635478987963,
      "grad_norm": 0.13303737450718892,
      "learning_rate": 3.3203847448811545e-05,
      "loss": 1.0785,
      "step": 10278
    },
    {
      "epoch": 0.7409356303611332,
      "grad_norm": 0.11383540381010578,
      "learning_rate": 3.318647441748055e-05,
      "loss": 0.9807,
      "step": 10279
    },
    {
      "epoch": 0.74100771282347,
      "grad_norm": 0.11357847658589339,
      "learning_rate": 3.3169105028084456e-05,
      "loss": 1.076,
      "step": 10280
    },
    {
      "epoch": 0.7410797952858069,
      "grad_norm": 0.11218489364676393,
      "learning_rate": 3.315173928156999e-05,
      "loss": 1.0915,
      "step": 10281
    },
    {
      "epoch": 0.7411518777481438,
      "grad_norm": 0.11022376112786478,
      "learning_rate": 3.313437717888379e-05,
      "loss": 1.0367,
      "step": 10282
    },
    {
      "epoch": 0.7412239602104808,
      "grad_norm": 0.1497443302483912,
      "learning_rate": 3.311701872097218e-05,
      "loss": 1.0699,
      "step": 10283
    },
    {
      "epoch": 0.7412960426728177,
      "grad_norm": 0.10731824883036954,
      "learning_rate": 3.309966390878149e-05,
      "loss": 1.0016,
      "step": 10284
    },
    {
      "epoch": 0.7413681251351546,
      "grad_norm": 0.12168113626158628,
      "learning_rate": 3.3082312743257574e-05,
      "loss": 1.0552,
      "step": 10285
    },
    {
      "epoch": 0.7414402075974915,
      "grad_norm": 0.10730478073163306,
      "learning_rate": 3.306496522534628e-05,
      "loss": 1.0427,
      "step": 10286
    },
    {
      "epoch": 0.7415122900598284,
      "grad_norm": 0.14426417944163566,
      "learning_rate": 3.3047621355993165e-05,
      "loss": 0.9782,
      "step": 10287
    },
    {
      "epoch": 0.7415843725221654,
      "grad_norm": 0.11614763559807413,
      "learning_rate": 3.3030281136143723e-05,
      "loss": 1.0125,
      "step": 10288
    },
    {
      "epoch": 0.7416564549845023,
      "grad_norm": 0.12876531038078726,
      "learning_rate": 3.3012944566743054e-05,
      "loss": 0.9327,
      "step": 10289
    },
    {
      "epoch": 0.7417285374468392,
      "grad_norm": 0.15017096283569373,
      "learning_rate": 3.29956116487362e-05,
      "loss": 1.027,
      "step": 10290
    },
    {
      "epoch": 0.7418006199091761,
      "grad_norm": 0.13652544998284005,
      "learning_rate": 3.297828238306794e-05,
      "loss": 1.046,
      "step": 10291
    },
    {
      "epoch": 0.741872702371513,
      "grad_norm": 0.12018996763463859,
      "learning_rate": 3.296095677068288e-05,
      "loss": 1.0383,
      "step": 10292
    },
    {
      "epoch": 0.74194478483385,
      "grad_norm": 0.24615023926809435,
      "learning_rate": 3.2943634812525435e-05,
      "loss": 0.9772,
      "step": 10293
    },
    {
      "epoch": 0.7420168672961869,
      "grad_norm": 0.1055073286470886,
      "learning_rate": 3.292631650953979e-05,
      "loss": 0.963,
      "step": 10294
    },
    {
      "epoch": 0.7420889497585238,
      "grad_norm": 0.11688656109586323,
      "learning_rate": 3.290900186266995e-05,
      "loss": 1.093,
      "step": 10295
    },
    {
      "epoch": 0.7421610322208607,
      "grad_norm": 0.20221906930369396,
      "learning_rate": 3.2891690872859736e-05,
      "loss": 1.0703,
      "step": 10296
    },
    {
      "epoch": 0.7422331146831975,
      "grad_norm": 0.1297005581870086,
      "learning_rate": 3.287438354105272e-05,
      "loss": 0.9537,
      "step": 10297
    },
    {
      "epoch": 0.7423051971455344,
      "grad_norm": 0.12521542101263722,
      "learning_rate": 3.285707986819234e-05,
      "loss": 1.0548,
      "step": 10298
    },
    {
      "epoch": 0.7423772796078714,
      "grad_norm": 0.11166172700354972,
      "learning_rate": 3.283977985522178e-05,
      "loss": 1.0171,
      "step": 10299
    },
    {
      "epoch": 0.7424493620702083,
      "grad_norm": 0.11632609080152718,
      "learning_rate": 3.2822483503084045e-05,
      "loss": 1.0614,
      "step": 10300
    },
    {
      "epoch": 0.7425214445325452,
      "grad_norm": 0.1091842479267784,
      "learning_rate": 3.2805190812721955e-05,
      "loss": 0.9871,
      "step": 10301
    },
    {
      "epoch": 0.7425935269948821,
      "grad_norm": 0.12003469566210521,
      "learning_rate": 3.278790178507809e-05,
      "loss": 1.0721,
      "step": 10302
    },
    {
      "epoch": 0.742665609457219,
      "grad_norm": 0.10753476917602549,
      "learning_rate": 3.277061642109489e-05,
      "loss": 0.9784,
      "step": 10303
    },
    {
      "epoch": 0.742737691919556,
      "grad_norm": 0.13369428992914476,
      "learning_rate": 3.275333472171455e-05,
      "loss": 0.9827,
      "step": 10304
    },
    {
      "epoch": 0.7428097743818929,
      "grad_norm": 0.13672097454332305,
      "learning_rate": 3.273605668787906e-05,
      "loss": 1.0941,
      "step": 10305
    },
    {
      "epoch": 0.7428818568442298,
      "grad_norm": 0.12411118178402908,
      "learning_rate": 3.271878232053025e-05,
      "loss": 1.0697,
      "step": 10306
    },
    {
      "epoch": 0.7429539393065667,
      "grad_norm": 0.11654143291293999,
      "learning_rate": 3.270151162060971e-05,
      "loss": 1.0203,
      "step": 10307
    },
    {
      "epoch": 0.7430260217689036,
      "grad_norm": 0.10823230533301484,
      "learning_rate": 3.2684244589058867e-05,
      "loss": 0.9845,
      "step": 10308
    },
    {
      "epoch": 0.7430981042312406,
      "grad_norm": 0.1230501219123405,
      "learning_rate": 3.26669812268189e-05,
      "loss": 1.0637,
      "step": 10309
    },
    {
      "epoch": 0.7431701866935775,
      "grad_norm": 0.16054664728008014,
      "learning_rate": 3.264972153483086e-05,
      "loss": 0.9997,
      "step": 10310
    },
    {
      "epoch": 0.7432422691559144,
      "grad_norm": 0.3229473153956571,
      "learning_rate": 3.263246551403552e-05,
      "loss": 1.0099,
      "step": 10311
    },
    {
      "epoch": 0.7433143516182513,
      "grad_norm": 0.11933680424712426,
      "learning_rate": 3.261521316537348e-05,
      "loss": 1.0166,
      "step": 10312
    },
    {
      "epoch": 0.7433864340805882,
      "grad_norm": 0.11775196951824612,
      "learning_rate": 3.259796448978522e-05,
      "loss": 1.0148,
      "step": 10313
    },
    {
      "epoch": 0.743458516542925,
      "grad_norm": 0.17484365441700736,
      "learning_rate": 3.258071948821081e-05,
      "loss": 1.0519,
      "step": 10314
    },
    {
      "epoch": 0.743530599005262,
      "grad_norm": 1.029653813658718,
      "learning_rate": 3.256347816159038e-05,
      "loss": 1.1369,
      "step": 10315
    },
    {
      "epoch": 0.7436026814675989,
      "grad_norm": 0.15425526747613857,
      "learning_rate": 3.2546240510863676e-05,
      "loss": 1.0065,
      "step": 10316
    },
    {
      "epoch": 0.7436747639299358,
      "grad_norm": 0.1209856442989299,
      "learning_rate": 3.2529006536970373e-05,
      "loss": 1.0313,
      "step": 10317
    },
    {
      "epoch": 0.7437468463922727,
      "grad_norm": 0.10559520384767575,
      "learning_rate": 3.2511776240849757e-05,
      "loss": 1.0173,
      "step": 10318
    },
    {
      "epoch": 0.7438189288546097,
      "grad_norm": 0.55630251271823,
      "learning_rate": 3.249454962344114e-05,
      "loss": 1.1084,
      "step": 10319
    },
    {
      "epoch": 0.7438910113169466,
      "grad_norm": 0.12407428476332842,
      "learning_rate": 3.2477326685683484e-05,
      "loss": 1.0347,
      "step": 10320
    },
    {
      "epoch": 0.7439630937792835,
      "grad_norm": 0.10747820230386287,
      "learning_rate": 3.246010742851564e-05,
      "loss": 1.0247,
      "step": 10321
    },
    {
      "epoch": 0.7440351762416204,
      "grad_norm": 0.11855670215320935,
      "learning_rate": 3.244289185287609e-05,
      "loss": 1.0241,
      "step": 10322
    },
    {
      "epoch": 0.7441072587039573,
      "grad_norm": 0.13123534938325618,
      "learning_rate": 3.242567995970336e-05,
      "loss": 1.0441,
      "step": 10323
    },
    {
      "epoch": 0.7441793411662943,
      "grad_norm": 0.1522782361846244,
      "learning_rate": 3.240847174993561e-05,
      "loss": 1.0164,
      "step": 10324
    },
    {
      "epoch": 0.7442514236286312,
      "grad_norm": 0.17415017845496913,
      "learning_rate": 3.2391267224510866e-05,
      "loss": 1.0382,
      "step": 10325
    },
    {
      "epoch": 0.7443235060909681,
      "grad_norm": 0.11785631045253894,
      "learning_rate": 3.2374066384366874e-05,
      "loss": 1.0028,
      "step": 10326
    },
    {
      "epoch": 0.744395588553305,
      "grad_norm": 0.11402184420352339,
      "learning_rate": 3.235686923044123e-05,
      "loss": 1.0016,
      "step": 10327
    },
    {
      "epoch": 0.7444676710156419,
      "grad_norm": 0.13559217395210987,
      "learning_rate": 3.23396757636714e-05,
      "loss": 1.0809,
      "step": 10328
    },
    {
      "epoch": 0.7445397534779788,
      "grad_norm": 0.14075896290092985,
      "learning_rate": 3.232248598499459e-05,
      "loss": 1.0816,
      "step": 10329
    },
    {
      "epoch": 0.7446118359403158,
      "grad_norm": 0.1423253081385631,
      "learning_rate": 3.230529989534771e-05,
      "loss": 1.0118,
      "step": 10330
    },
    {
      "epoch": 0.7446839184026527,
      "grad_norm": 0.1253864326529032,
      "learning_rate": 3.228811749566757e-05,
      "loss": 1.0112,
      "step": 10331
    },
    {
      "epoch": 0.7447560008649895,
      "grad_norm": 0.1170461364946704,
      "learning_rate": 3.227093878689082e-05,
      "loss": 1.02,
      "step": 10332
    },
    {
      "epoch": 0.7448280833273264,
      "grad_norm": 0.10664690217014121,
      "learning_rate": 3.225376376995387e-05,
      "loss": 0.953,
      "step": 10333
    },
    {
      "epoch": 0.7449001657896633,
      "grad_norm": 0.24238204026493357,
      "learning_rate": 3.223659244579284e-05,
      "loss": 1.0646,
      "step": 10334
    },
    {
      "epoch": 0.7449722482520003,
      "grad_norm": 0.11389890867294968,
      "learning_rate": 3.2219424815343735e-05,
      "loss": 0.9748,
      "step": 10335
    },
    {
      "epoch": 0.7450443307143372,
      "grad_norm": 0.1662198367173383,
      "learning_rate": 3.2202260879542335e-05,
      "loss": 1.0577,
      "step": 10336
    },
    {
      "epoch": 0.7451164131766741,
      "grad_norm": 0.10982421044282233,
      "learning_rate": 3.218510063932432e-05,
      "loss": 1.0446,
      "step": 10337
    },
    {
      "epoch": 0.745188495639011,
      "grad_norm": 0.11648435072998699,
      "learning_rate": 3.216794409562498e-05,
      "loss": 1.0183,
      "step": 10338
    },
    {
      "epoch": 0.7452605781013479,
      "grad_norm": 0.1293367572427684,
      "learning_rate": 3.2150791249379516e-05,
      "loss": 1.1091,
      "step": 10339
    },
    {
      "epoch": 0.7453326605636849,
      "grad_norm": 0.11628393884441877,
      "learning_rate": 3.213364210152291e-05,
      "loss": 1.0317,
      "step": 10340
    },
    {
      "epoch": 0.7454047430260218,
      "grad_norm": 0.1425812624016755,
      "learning_rate": 3.211649665299001e-05,
      "loss": 1.0316,
      "step": 10341
    },
    {
      "epoch": 0.7454768254883587,
      "grad_norm": 0.11955824804548154,
      "learning_rate": 3.209935490471533e-05,
      "loss": 1.0966,
      "step": 10342
    },
    {
      "epoch": 0.7455489079506956,
      "grad_norm": 0.14099767819744669,
      "learning_rate": 3.208221685763325e-05,
      "loss": 0.9993,
      "step": 10343
    },
    {
      "epoch": 0.7456209904130325,
      "grad_norm": 0.11250447501250044,
      "learning_rate": 3.2065082512677936e-05,
      "loss": 0.9675,
      "step": 10344
    },
    {
      "epoch": 0.7456930728753695,
      "grad_norm": 0.12919276738007535,
      "learning_rate": 3.204795187078344e-05,
      "loss": 1.0605,
      "step": 10345
    },
    {
      "epoch": 0.7457651553377064,
      "grad_norm": 0.11318597432654647,
      "learning_rate": 3.203082493288346e-05,
      "loss": 1.1251,
      "step": 10346
    },
    {
      "epoch": 0.7458372378000433,
      "grad_norm": 0.11543246156704451,
      "learning_rate": 3.201370169991159e-05,
      "loss": 1.0266,
      "step": 10347
    },
    {
      "epoch": 0.7459093202623802,
      "grad_norm": 0.11222336479498868,
      "learning_rate": 3.19965821728012e-05,
      "loss": 1.025,
      "step": 10348
    },
    {
      "epoch": 0.745981402724717,
      "grad_norm": 0.12152068047888957,
      "learning_rate": 3.1979466352485464e-05,
      "loss": 1.0883,
      "step": 10349
    },
    {
      "epoch": 0.7460534851870539,
      "grad_norm": 0.15482127408890609,
      "learning_rate": 3.196235423989734e-05,
      "loss": 0.9756,
      "step": 10350
    },
    {
      "epoch": 0.7461255676493909,
      "grad_norm": 0.4778561394589877,
      "learning_rate": 3.194524583596958e-05,
      "loss": 1.0929,
      "step": 10351
    },
    {
      "epoch": 0.7461976501117278,
      "grad_norm": 0.13761463670922552,
      "learning_rate": 3.1928141141634773e-05,
      "loss": 1.0808,
      "step": 10352
    },
    {
      "epoch": 0.7462697325740647,
      "grad_norm": 0.1279369430271194,
      "learning_rate": 3.191104015782525e-05,
      "loss": 0.9729,
      "step": 10353
    },
    {
      "epoch": 0.7463418150364016,
      "grad_norm": 0.11325061304865496,
      "learning_rate": 3.1893942885473184e-05,
      "loss": 1.0521,
      "step": 10354
    },
    {
      "epoch": 0.7464138974987385,
      "grad_norm": 0.12111225068070725,
      "learning_rate": 3.187684932551053e-05,
      "loss": 0.9925,
      "step": 10355
    },
    {
      "epoch": 0.7464859799610755,
      "grad_norm": 0.11150258939181386,
      "learning_rate": 3.1859759478869024e-05,
      "loss": 1.0337,
      "step": 10356
    },
    {
      "epoch": 0.7465580624234124,
      "grad_norm": 0.11819745262740293,
      "learning_rate": 3.1842673346480235e-05,
      "loss": 0.9916,
      "step": 10357
    },
    {
      "epoch": 0.7466301448857493,
      "grad_norm": 0.131533391008219,
      "learning_rate": 3.18255909292755e-05,
      "loss": 1.0804,
      "step": 10358
    },
    {
      "epoch": 0.7467022273480862,
      "grad_norm": 0.11697035184630614,
      "learning_rate": 3.180851222818596e-05,
      "loss": 0.985,
      "step": 10359
    },
    {
      "epoch": 0.7467743098104231,
      "grad_norm": 0.11808698623438695,
      "learning_rate": 3.179143724414257e-05,
      "loss": 1.0436,
      "step": 10360
    },
    {
      "epoch": 0.7468463922727601,
      "grad_norm": 0.12675143276110495,
      "learning_rate": 3.177436597807606e-05,
      "loss": 1.0455,
      "step": 10361
    },
    {
      "epoch": 0.746918474735097,
      "grad_norm": 0.14693958181589917,
      "learning_rate": 3.175729843091696e-05,
      "loss": 1.083,
      "step": 10362
    },
    {
      "epoch": 0.7469905571974339,
      "grad_norm": 0.11724080410558793,
      "learning_rate": 3.174023460359562e-05,
      "loss": 1.0463,
      "step": 10363
    },
    {
      "epoch": 0.7470626396597708,
      "grad_norm": 0.18166344443432442,
      "learning_rate": 3.172317449704216e-05,
      "loss": 1.044,
      "step": 10364
    },
    {
      "epoch": 0.7471347221221077,
      "grad_norm": 0.14136958612222908,
      "learning_rate": 3.170611811218651e-05,
      "loss": 1.0265,
      "step": 10365
    },
    {
      "epoch": 0.7472068045844446,
      "grad_norm": 0.11225137051243969,
      "learning_rate": 3.16890654499584e-05,
      "loss": 1.0729,
      "step": 10366
    },
    {
      "epoch": 0.7472788870467815,
      "grad_norm": 0.11120765884813096,
      "learning_rate": 3.167201651128735e-05,
      "loss": 0.9418,
      "step": 10367
    },
    {
      "epoch": 0.7473509695091184,
      "grad_norm": 0.14555965510159447,
      "learning_rate": 3.165497129710269e-05,
      "loss": 1.0345,
      "step": 10368
    },
    {
      "epoch": 0.7474230519714553,
      "grad_norm": 0.9980727902736328,
      "learning_rate": 3.163792980833351e-05,
      "loss": 1.0394,
      "step": 10369
    },
    {
      "epoch": 0.7474951344337922,
      "grad_norm": 0.1510477236132484,
      "learning_rate": 3.162089204590878e-05,
      "loss": 1.0248,
      "step": 10370
    },
    {
      "epoch": 0.7475672168961291,
      "grad_norm": 0.12479455476744857,
      "learning_rate": 3.1603858010757104e-05,
      "loss": 1.0474,
      "step": 10371
    },
    {
      "epoch": 0.7476392993584661,
      "grad_norm": 0.13110174980922376,
      "learning_rate": 3.158682770380709e-05,
      "loss": 1.0307,
      "step": 10372
    },
    {
      "epoch": 0.747711381820803,
      "grad_norm": 0.12457449499030797,
      "learning_rate": 3.156980112598701e-05,
      "loss": 0.9708,
      "step": 10373
    },
    {
      "epoch": 0.7477834642831399,
      "grad_norm": 0.12954020645493816,
      "learning_rate": 3.155277827822498e-05,
      "loss": 1.0302,
      "step": 10374
    },
    {
      "epoch": 0.7478555467454768,
      "grad_norm": 0.11901003379469945,
      "learning_rate": 3.153575916144883e-05,
      "loss": 1.0407,
      "step": 10375
    },
    {
      "epoch": 0.7479276292078137,
      "grad_norm": 0.11987797547436828,
      "learning_rate": 3.151874377658633e-05,
      "loss": 1.0444,
      "step": 10376
    },
    {
      "epoch": 0.7479997116701507,
      "grad_norm": 0.16600286181811302,
      "learning_rate": 3.1501732124564954e-05,
      "loss": 0.9908,
      "step": 10377
    },
    {
      "epoch": 0.7480717941324876,
      "grad_norm": 0.129777213126852,
      "learning_rate": 3.148472420631201e-05,
      "loss": 1.1271,
      "step": 10378
    },
    {
      "epoch": 0.7481438765948245,
      "grad_norm": 0.13762982509091945,
      "learning_rate": 3.146772002275452e-05,
      "loss": 0.9509,
      "step": 10379
    },
    {
      "epoch": 0.7482159590571614,
      "grad_norm": 0.14295519041057655,
      "learning_rate": 3.145071957481935e-05,
      "loss": 1.1637,
      "step": 10380
    },
    {
      "epoch": 0.7482880415194983,
      "grad_norm": 0.13138724703483431,
      "learning_rate": 3.143372286343327e-05,
      "loss": 0.9936,
      "step": 10381
    },
    {
      "epoch": 0.7483601239818353,
      "grad_norm": 0.12625811892770286,
      "learning_rate": 3.1416729889522735e-05,
      "loss": 1.0497,
      "step": 10382
    },
    {
      "epoch": 0.7484322064441722,
      "grad_norm": 0.12785860835143328,
      "learning_rate": 3.139974065401395e-05,
      "loss": 0.9676,
      "step": 10383
    },
    {
      "epoch": 0.748504288906509,
      "grad_norm": 0.24455257116436516,
      "learning_rate": 3.1382755157832986e-05,
      "loss": 1.1071,
      "step": 10384
    },
    {
      "epoch": 0.7485763713688459,
      "grad_norm": 0.11804374688940537,
      "learning_rate": 3.136577340190576e-05,
      "loss": 0.9729,
      "step": 10385
    },
    {
      "epoch": 0.7486484538311828,
      "grad_norm": 0.12570649929552885,
      "learning_rate": 3.134879538715794e-05,
      "loss": 0.9874,
      "step": 10386
    },
    {
      "epoch": 0.7487205362935198,
      "grad_norm": 0.11350899007261236,
      "learning_rate": 3.1331821114514905e-05,
      "loss": 1.0261,
      "step": 10387
    },
    {
      "epoch": 0.7487926187558567,
      "grad_norm": 0.1351731921834128,
      "learning_rate": 3.131485058490191e-05,
      "loss": 1.027,
      "step": 10388
    },
    {
      "epoch": 0.7488647012181936,
      "grad_norm": 0.1351172876747151,
      "learning_rate": 3.129788379924408e-05,
      "loss": 1.0261,
      "step": 10389
    },
    {
      "epoch": 0.7489367836805305,
      "grad_norm": 0.1138197037652048,
      "learning_rate": 3.128092075846623e-05,
      "loss": 1.0309,
      "step": 10390
    },
    {
      "epoch": 0.7490088661428674,
      "grad_norm": 0.13669275609324066,
      "learning_rate": 3.126396146349294e-05,
      "loss": 1.0213,
      "step": 10391
    },
    {
      "epoch": 0.7490809486052044,
      "grad_norm": 0.1174338034308873,
      "learning_rate": 3.1247005915248686e-05,
      "loss": 0.9883,
      "step": 10392
    },
    {
      "epoch": 0.7491530310675413,
      "grad_norm": 0.13079701385961742,
      "learning_rate": 3.123005411465766e-05,
      "loss": 1.0473,
      "step": 10393
    },
    {
      "epoch": 0.7492251135298782,
      "grad_norm": 0.12409114038743779,
      "learning_rate": 3.121310606264398e-05,
      "loss": 1.0466,
      "step": 10394
    },
    {
      "epoch": 0.7492971959922151,
      "grad_norm": 1.7132269349542717,
      "learning_rate": 3.119616176013137e-05,
      "loss": 1.0767,
      "step": 10395
    },
    {
      "epoch": 0.749369278454552,
      "grad_norm": 0.21472727051587112,
      "learning_rate": 3.117922120804349e-05,
      "loss": 0.9869,
      "step": 10396
    },
    {
      "epoch": 0.749441360916889,
      "grad_norm": 0.1256136220200032,
      "learning_rate": 3.11622844073037e-05,
      "loss": 1.1012,
      "step": 10397
    },
    {
      "epoch": 0.7495134433792259,
      "grad_norm": 0.1244310863214964,
      "learning_rate": 3.114535135883532e-05,
      "loss": 1.0725,
      "step": 10398
    },
    {
      "epoch": 0.7495855258415628,
      "grad_norm": 0.1198882587978725,
      "learning_rate": 3.112842206356125e-05,
      "loss": 1.0905,
      "step": 10399
    },
    {
      "epoch": 0.7496576083038997,
      "grad_norm": 0.12239978490891353,
      "learning_rate": 3.1111496522404315e-05,
      "loss": 1.0077,
      "step": 10400
    },
    {
      "epoch": 0.7497296907662365,
      "grad_norm": 0.11894098346817202,
      "learning_rate": 3.109457473628712e-05,
      "loss": 0.9633,
      "step": 10401
    },
    {
      "epoch": 0.7498017732285734,
      "grad_norm": 0.12233726264680796,
      "learning_rate": 3.1077656706132054e-05,
      "loss": 1.0264,
      "step": 10402
    },
    {
      "epoch": 0.7498738556909104,
      "grad_norm": 0.12429893033405476,
      "learning_rate": 3.106074243286128e-05,
      "loss": 0.9816,
      "step": 10403
    },
    {
      "epoch": 0.7499459381532473,
      "grad_norm": 0.12537064983656918,
      "learning_rate": 3.1043831917396795e-05,
      "loss": 1.0119,
      "step": 10404
    },
    {
      "epoch": 0.7500180206155842,
      "grad_norm": 0.11459471345707833,
      "learning_rate": 3.102692516066037e-05,
      "loss": 1.1049,
      "step": 10405
    },
    {
      "epoch": 0.7500901030779211,
      "grad_norm": 0.13201645792264846,
      "learning_rate": 3.1010022163573585e-05,
      "loss": 1.0581,
      "step": 10406
    },
    {
      "epoch": 0.750162185540258,
      "grad_norm": 0.21326333685629087,
      "learning_rate": 3.099312292705778e-05,
      "loss": 1.0886,
      "step": 10407
    },
    {
      "epoch": 0.750234268002595,
      "grad_norm": 0.12484452543715767,
      "learning_rate": 3.097622745203413e-05,
      "loss": 1.0168,
      "step": 10408
    },
    {
      "epoch": 0.7503063504649319,
      "grad_norm": 0.13521148955147652,
      "learning_rate": 3.09593357394236e-05,
      "loss": 1.0403,
      "step": 10409
    },
    {
      "epoch": 0.7503784329272688,
      "grad_norm": 0.10915310645853493,
      "learning_rate": 3.0942447790146925e-05,
      "loss": 1.0446,
      "step": 10410
    },
    {
      "epoch": 0.7504505153896057,
      "grad_norm": 0.15983593511928115,
      "learning_rate": 3.092556360512464e-05,
      "loss": 1.0396,
      "step": 10411
    },
    {
      "epoch": 0.7505225978519426,
      "grad_norm": 0.11571786943448002,
      "learning_rate": 3.090868318527711e-05,
      "loss": 0.9673,
      "step": 10412
    },
    {
      "epoch": 0.7505946803142796,
      "grad_norm": 0.11629234559671142,
      "learning_rate": 3.089180653152445e-05,
      "loss": 0.9977,
      "step": 10413
    },
    {
      "epoch": 0.7506667627766165,
      "grad_norm": 0.1079532064379124,
      "learning_rate": 3.087493364478659e-05,
      "loss": 0.9471,
      "step": 10414
    },
    {
      "epoch": 0.7507388452389534,
      "grad_norm": 0.14105426074380462,
      "learning_rate": 3.085806452598325e-05,
      "loss": 1.0545,
      "step": 10415
    },
    {
      "epoch": 0.7508109277012903,
      "grad_norm": 0.1260222421588242,
      "learning_rate": 3.0841199176033974e-05,
      "loss": 1.045,
      "step": 10416
    },
    {
      "epoch": 0.7508830101636272,
      "grad_norm": 0.15805092546064067,
      "learning_rate": 3.082433759585804e-05,
      "loss": 0.9692,
      "step": 10417
    },
    {
      "epoch": 0.750955092625964,
      "grad_norm": 0.13598004943595055,
      "learning_rate": 3.080747978637457e-05,
      "loss": 1.0535,
      "step": 10418
    },
    {
      "epoch": 0.751027175088301,
      "grad_norm": 0.11487477439295177,
      "learning_rate": 3.0790625748502475e-05,
      "loss": 1.0182,
      "step": 10419
    },
    {
      "epoch": 0.7510992575506379,
      "grad_norm": 0.13907246205853865,
      "learning_rate": 3.077377548316044e-05,
      "loss": 1.0289,
      "step": 10420
    },
    {
      "epoch": 0.7511713400129748,
      "grad_norm": 0.1422588529927734,
      "learning_rate": 3.075692899126695e-05,
      "loss": 0.9647,
      "step": 10421
    },
    {
      "epoch": 0.7512434224753117,
      "grad_norm": 0.11402884242897944,
      "learning_rate": 3.0740086273740295e-05,
      "loss": 1.029,
      "step": 10422
    },
    {
      "epoch": 0.7513155049376486,
      "grad_norm": 0.12285876464625298,
      "learning_rate": 3.072324733149861e-05,
      "loss": 0.9955,
      "step": 10423
    },
    {
      "epoch": 0.7513875873999856,
      "grad_norm": 0.11766864972429368,
      "learning_rate": 3.070641216545962e-05,
      "loss": 1.0363,
      "step": 10424
    },
    {
      "epoch": 0.7514596698623225,
      "grad_norm": 0.13478340082262122,
      "learning_rate": 3.068958077654113e-05,
      "loss": 1.0301,
      "step": 10425
    },
    {
      "epoch": 0.7515317523246594,
      "grad_norm": 0.10855740838369368,
      "learning_rate": 3.067275316566055e-05,
      "loss": 1.0213,
      "step": 10426
    },
    {
      "epoch": 0.7516038347869963,
      "grad_norm": 0.1363111586175716,
      "learning_rate": 3.065592933373518e-05,
      "loss": 1.1015,
      "step": 10427
    },
    {
      "epoch": 0.7516759172493332,
      "grad_norm": 0.1246650394408507,
      "learning_rate": 3.0639109281681966e-05,
      "loss": 1.035,
      "step": 10428
    },
    {
      "epoch": 0.7517479997116702,
      "grad_norm": 0.12610517845409533,
      "learning_rate": 3.062229301041785e-05,
      "loss": 1.0743,
      "step": 10429
    },
    {
      "epoch": 0.7518200821740071,
      "grad_norm": 0.2101051997375953,
      "learning_rate": 3.0605480520859433e-05,
      "loss": 1.0483,
      "step": 10430
    },
    {
      "epoch": 0.751892164636344,
      "grad_norm": 0.12508245609014845,
      "learning_rate": 3.058867181392319e-05,
      "loss": 1.0611,
      "step": 10431
    },
    {
      "epoch": 0.7519642470986809,
      "grad_norm": 0.14651381516479386,
      "learning_rate": 3.057186689052523e-05,
      "loss": 0.9487,
      "step": 10432
    },
    {
      "epoch": 0.7520363295610178,
      "grad_norm": 0.12033316045808251,
      "learning_rate": 3.055506575158168e-05,
      "loss": 1.0562,
      "step": 10433
    },
    {
      "epoch": 0.7521084120233548,
      "grad_norm": 0.13572946057126176,
      "learning_rate": 3.053826839800834e-05,
      "loss": 0.9385,
      "step": 10434
    },
    {
      "epoch": 0.7521804944856917,
      "grad_norm": 0.13934652451143834,
      "learning_rate": 3.052147483072078e-05,
      "loss": 1.041,
      "step": 10435
    },
    {
      "epoch": 0.7522525769480285,
      "grad_norm": 0.11717778391995733,
      "learning_rate": 3.050468505063441e-05,
      "loss": 0.9744,
      "step": 10436
    },
    {
      "epoch": 0.7523246594103654,
      "grad_norm": 0.1322830953810172,
      "learning_rate": 3.0487899058664395e-05,
      "loss": 1.088,
      "step": 10437
    },
    {
      "epoch": 0.7523967418727023,
      "grad_norm": 0.1348860009075859,
      "learning_rate": 3.0471116855725813e-05,
      "loss": 1.0241,
      "step": 10438
    },
    {
      "epoch": 0.7524688243350393,
      "grad_norm": 0.12239936867255266,
      "learning_rate": 3.0454338442733355e-05,
      "loss": 1.0818,
      "step": 10439
    },
    {
      "epoch": 0.7525409067973762,
      "grad_norm": 0.1162204996405431,
      "learning_rate": 3.0437563820601633e-05,
      "loss": 0.9806,
      "step": 10440
    },
    {
      "epoch": 0.7526129892597131,
      "grad_norm": 0.14514176200412876,
      "learning_rate": 3.042079299024496e-05,
      "loss": 1.0001,
      "step": 10441
    },
    {
      "epoch": 0.75268507172205,
      "grad_norm": 0.14533683138107326,
      "learning_rate": 3.040402595257762e-05,
      "loss": 1.0264,
      "step": 10442
    },
    {
      "epoch": 0.7527571541843869,
      "grad_norm": 0.1634162145926644,
      "learning_rate": 3.0387262708513454e-05,
      "loss": 0.987,
      "step": 10443
    },
    {
      "epoch": 0.7528292366467239,
      "grad_norm": 0.1315347730543154,
      "learning_rate": 3.0370503258966244e-05,
      "loss": 0.991,
      "step": 10444
    },
    {
      "epoch": 0.7529013191090608,
      "grad_norm": 0.12401438806978689,
      "learning_rate": 3.0353747604849527e-05,
      "loss": 1.0224,
      "step": 10445
    },
    {
      "epoch": 0.7529734015713977,
      "grad_norm": 0.13703360522935334,
      "learning_rate": 3.0336995747076636e-05,
      "loss": 0.9934,
      "step": 10446
    },
    {
      "epoch": 0.7530454840337346,
      "grad_norm": 0.14669303268946485,
      "learning_rate": 3.03202476865607e-05,
      "loss": 1.0165,
      "step": 10447
    },
    {
      "epoch": 0.7531175664960715,
      "grad_norm": 0.12018618666272918,
      "learning_rate": 3.0303503424214643e-05,
      "loss": 1.0114,
      "step": 10448
    },
    {
      "epoch": 0.7531896489584085,
      "grad_norm": 0.13622549739951914,
      "learning_rate": 3.0286762960951166e-05,
      "loss": 1.0569,
      "step": 10449
    },
    {
      "epoch": 0.7532617314207454,
      "grad_norm": 0.14293709523457923,
      "learning_rate": 3.0270026297682785e-05,
      "loss": 0.9814,
      "step": 10450
    },
    {
      "epoch": 0.7533338138830823,
      "grad_norm": 0.13373015847951322,
      "learning_rate": 3.0253293435321793e-05,
      "loss": 1.0066,
      "step": 10451
    },
    {
      "epoch": 0.7534058963454192,
      "grad_norm": 0.11520746950182639,
      "learning_rate": 3.0236564374780284e-05,
      "loss": 1.0053,
      "step": 10452
    },
    {
      "epoch": 0.753477978807756,
      "grad_norm": 0.13347254802281153,
      "learning_rate": 3.021983911697013e-05,
      "loss": 1.0303,
      "step": 10453
    },
    {
      "epoch": 0.7535500612700929,
      "grad_norm": 0.126159903625069,
      "learning_rate": 3.0203117662803016e-05,
      "loss": 0.9895,
      "step": 10454
    },
    {
      "epoch": 0.7536221437324299,
      "grad_norm": 0.12822822399115158,
      "learning_rate": 3.018640001319042e-05,
      "loss": 0.9699,
      "step": 10455
    },
    {
      "epoch": 0.7536942261947668,
      "grad_norm": 0.11772221084194537,
      "learning_rate": 3.0169686169043587e-05,
      "loss": 1.0431,
      "step": 10456
    },
    {
      "epoch": 0.7537663086571037,
      "grad_norm": 0.12098621132815054,
      "learning_rate": 3.0152976131273568e-05,
      "loss": 0.9498,
      "step": 10457
    },
    {
      "epoch": 0.7538383911194406,
      "grad_norm": 0.13417249902136308,
      "learning_rate": 3.0136269900791235e-05,
      "loss": 1.0648,
      "step": 10458
    },
    {
      "epoch": 0.7539104735817775,
      "grad_norm": 0.829740509626358,
      "learning_rate": 3.0119567478507206e-05,
      "loss": 1.0119,
      "step": 10459
    },
    {
      "epoch": 0.7539825560441145,
      "grad_norm": 0.11547255915417765,
      "learning_rate": 3.0102868865331914e-05,
      "loss": 0.9987,
      "step": 10460
    },
    {
      "epoch": 0.7540546385064514,
      "grad_norm": 0.1245260324587215,
      "learning_rate": 3.008617406217559e-05,
      "loss": 0.9997,
      "step": 10461
    },
    {
      "epoch": 0.7541267209687883,
      "grad_norm": 0.11764904280465012,
      "learning_rate": 3.0069483069948246e-05,
      "loss": 1.0296,
      "step": 10462
    },
    {
      "epoch": 0.7541988034311252,
      "grad_norm": 0.11539597366616139,
      "learning_rate": 3.0052795889559694e-05,
      "loss": 0.986,
      "step": 10463
    },
    {
      "epoch": 0.7542708858934621,
      "grad_norm": 0.11702605085413934,
      "learning_rate": 3.0036112521919525e-05,
      "loss": 1.0052,
      "step": 10464
    },
    {
      "epoch": 0.7543429683557991,
      "grad_norm": 0.11934647545489234,
      "learning_rate": 3.0019432967937143e-05,
      "loss": 1.0175,
      "step": 10465
    },
    {
      "epoch": 0.754415050818136,
      "grad_norm": 0.1274222281066185,
      "learning_rate": 3.0002757228521726e-05,
      "loss": 1.0589,
      "step": 10466
    },
    {
      "epoch": 0.7544871332804729,
      "grad_norm": 0.11129332946137782,
      "learning_rate": 2.998608530458228e-05,
      "loss": 1.0609,
      "step": 10467
    },
    {
      "epoch": 0.7545592157428098,
      "grad_norm": 0.11552400061722669,
      "learning_rate": 2.9969417197027495e-05,
      "loss": 1.0575,
      "step": 10468
    },
    {
      "epoch": 0.7546312982051467,
      "grad_norm": 0.14231623389768855,
      "learning_rate": 2.9952752906766013e-05,
      "loss": 1.0417,
      "step": 10469
    },
    {
      "epoch": 0.7547033806674835,
      "grad_norm": 0.17494508798876468,
      "learning_rate": 2.9936092434706153e-05,
      "loss": 0.8989,
      "step": 10470
    },
    {
      "epoch": 0.7547754631298205,
      "grad_norm": 0.11634906502920052,
      "learning_rate": 2.99194357817561e-05,
      "loss": 0.9683,
      "step": 10471
    },
    {
      "epoch": 0.7548475455921574,
      "grad_norm": 0.11649976914700551,
      "learning_rate": 2.9902782948823694e-05,
      "loss": 1.0012,
      "step": 10472
    },
    {
      "epoch": 0.7549196280544943,
      "grad_norm": 0.1401332402432925,
      "learning_rate": 2.988613393681675e-05,
      "loss": 1.0712,
      "step": 10473
    },
    {
      "epoch": 0.7549917105168312,
      "grad_norm": 0.12479523228188319,
      "learning_rate": 2.986948874664276e-05,
      "loss": 0.9565,
      "step": 10474
    },
    {
      "epoch": 0.7550637929791681,
      "grad_norm": 0.12929071267762585,
      "learning_rate": 2.985284737920908e-05,
      "loss": 1.0121,
      "step": 10475
    },
    {
      "epoch": 0.7551358754415051,
      "grad_norm": 0.11335569839725534,
      "learning_rate": 2.9836209835422713e-05,
      "loss": 1.1165,
      "step": 10476
    },
    {
      "epoch": 0.755207957903842,
      "grad_norm": 0.3199478407847546,
      "learning_rate": 2.981957611619063e-05,
      "loss": 1.1102,
      "step": 10477
    },
    {
      "epoch": 0.7552800403661789,
      "grad_norm": 0.11231923350035464,
      "learning_rate": 2.9802946222419513e-05,
      "loss": 1.0219,
      "step": 10478
    },
    {
      "epoch": 0.7553521228285158,
      "grad_norm": 0.10606757831534448,
      "learning_rate": 2.9786320155015857e-05,
      "loss": 1.0307,
      "step": 10479
    },
    {
      "epoch": 0.7554242052908527,
      "grad_norm": 0.12978638390397354,
      "learning_rate": 2.9769697914885862e-05,
      "loss": 1.0184,
      "step": 10480
    },
    {
      "epoch": 0.7554962877531897,
      "grad_norm": 0.12953204538313282,
      "learning_rate": 2.975307950293561e-05,
      "loss": 1.0419,
      "step": 10481
    },
    {
      "epoch": 0.7555683702155266,
      "grad_norm": 0.1193914479405358,
      "learning_rate": 2.9736464920070993e-05,
      "loss": 0.9761,
      "step": 10482
    },
    {
      "epoch": 0.7556404526778635,
      "grad_norm": 0.14496504213482567,
      "learning_rate": 2.971985416719767e-05,
      "loss": 1.0165,
      "step": 10483
    },
    {
      "epoch": 0.7557125351402004,
      "grad_norm": 0.12063008012763195,
      "learning_rate": 2.970324724522101e-05,
      "loss": 1.0628,
      "step": 10484
    },
    {
      "epoch": 0.7557846176025373,
      "grad_norm": 0.13319886582238838,
      "learning_rate": 2.968664415504623e-05,
      "loss": 1.0372,
      "step": 10485
    },
    {
      "epoch": 0.7558567000648743,
      "grad_norm": 0.23559562510289506,
      "learning_rate": 2.9670044897578418e-05,
      "loss": 1.0859,
      "step": 10486
    },
    {
      "epoch": 0.7559287825272112,
      "grad_norm": 0.11586974830056576,
      "learning_rate": 2.9653449473722383e-05,
      "loss": 1.0041,
      "step": 10487
    },
    {
      "epoch": 0.756000864989548,
      "grad_norm": 0.12379305335027438,
      "learning_rate": 2.963685788438266e-05,
      "loss": 1.0625,
      "step": 10488
    },
    {
      "epoch": 0.7560729474518849,
      "grad_norm": 0.45382614151182166,
      "learning_rate": 2.9620270130463667e-05,
      "loss": 1.0932,
      "step": 10489
    },
    {
      "epoch": 0.7561450299142218,
      "grad_norm": 0.12831908020490412,
      "learning_rate": 2.9603686212869552e-05,
      "loss": 1.0775,
      "step": 10490
    },
    {
      "epoch": 0.7562171123765588,
      "grad_norm": 0.12739339899984867,
      "learning_rate": 2.9587106132504393e-05,
      "loss": 1.0319,
      "step": 10491
    },
    {
      "epoch": 0.7562891948388957,
      "grad_norm": 0.1167341550700186,
      "learning_rate": 2.9570529890271835e-05,
      "loss": 1.0591,
      "step": 10492
    },
    {
      "epoch": 0.7563612773012326,
      "grad_norm": 0.12110584674688948,
      "learning_rate": 2.955395748707549e-05,
      "loss": 1.0153,
      "step": 10493
    },
    {
      "epoch": 0.7564333597635695,
      "grad_norm": 0.15137680054202984,
      "learning_rate": 2.9537388923818656e-05,
      "loss": 1.0383,
      "step": 10494
    },
    {
      "epoch": 0.7565054422259064,
      "grad_norm": 0.13464531498651217,
      "learning_rate": 2.9520824201404573e-05,
      "loss": 1.0749,
      "step": 10495
    },
    {
      "epoch": 0.7565775246882434,
      "grad_norm": 0.12974954518255136,
      "learning_rate": 2.9504263320736046e-05,
      "loss": 1.0874,
      "step": 10496
    },
    {
      "epoch": 0.7566496071505803,
      "grad_norm": 0.13136114490102313,
      "learning_rate": 2.9487706282715856e-05,
      "loss": 1.0827,
      "step": 10497
    },
    {
      "epoch": 0.7567216896129172,
      "grad_norm": 1.6429274191364,
      "learning_rate": 2.9471153088246462e-05,
      "loss": 1.196,
      "step": 10498
    },
    {
      "epoch": 0.7567937720752541,
      "grad_norm": 0.14173256474406065,
      "learning_rate": 2.9454603738230256e-05,
      "loss": 1.0259,
      "step": 10499
    },
    {
      "epoch": 0.756865854537591,
      "grad_norm": 0.1120511675901732,
      "learning_rate": 2.943805823356922e-05,
      "loss": 0.9812,
      "step": 10500
    },
    {
      "epoch": 0.756937936999928,
      "grad_norm": 0.11485942240917754,
      "learning_rate": 2.9421516575165288e-05,
      "loss": 1.0152,
      "step": 10501
    },
    {
      "epoch": 0.7570100194622649,
      "grad_norm": 0.14163229095623323,
      "learning_rate": 2.940497876392011e-05,
      "loss": 0.9967,
      "step": 10502
    },
    {
      "epoch": 0.7570821019246018,
      "grad_norm": 0.1483580400794405,
      "learning_rate": 2.938844480073516e-05,
      "loss": 1.052,
      "step": 10503
    },
    {
      "epoch": 0.7571541843869387,
      "grad_norm": 0.5682788529729942,
      "learning_rate": 2.937191468651167e-05,
      "loss": 1.0606,
      "step": 10504
    },
    {
      "epoch": 0.7572262668492755,
      "grad_norm": 0.9786828028968217,
      "learning_rate": 2.9355388422150688e-05,
      "loss": 0.9725,
      "step": 10505
    },
    {
      "epoch": 0.7572983493116124,
      "grad_norm": 0.14152754456063518,
      "learning_rate": 2.9338866008553035e-05,
      "loss": 0.9798,
      "step": 10506
    },
    {
      "epoch": 0.7573704317739494,
      "grad_norm": 0.16566633566079625,
      "learning_rate": 2.9322347446619345e-05,
      "loss": 1.0149,
      "step": 10507
    },
    {
      "epoch": 0.7574425142362863,
      "grad_norm": 0.44273500923512626,
      "learning_rate": 2.9305832737250017e-05,
      "loss": 1.0106,
      "step": 10508
    },
    {
      "epoch": 0.7575145966986232,
      "grad_norm": 0.13492396228812592,
      "learning_rate": 2.9289321881345254e-05,
      "loss": 1.0265,
      "step": 10509
    },
    {
      "epoch": 0.7575866791609601,
      "grad_norm": 0.14121469741061962,
      "learning_rate": 2.9272814879805045e-05,
      "loss": 1.0563,
      "step": 10510
    },
    {
      "epoch": 0.757658761623297,
      "grad_norm": 0.12822980411232682,
      "learning_rate": 2.9256311733529163e-05,
      "loss": 1.0934,
      "step": 10511
    },
    {
      "epoch": 0.757730844085634,
      "grad_norm": 0.13584500777714262,
      "learning_rate": 2.923981244341719e-05,
      "loss": 1.1099,
      "step": 10512
    },
    {
      "epoch": 0.7578029265479709,
      "grad_norm": 0.12394365744538854,
      "learning_rate": 2.9223317010368478e-05,
      "loss": 1.0611,
      "step": 10513
    },
    {
      "epoch": 0.7578750090103078,
      "grad_norm": 0.13082471800882428,
      "learning_rate": 2.920682543528217e-05,
      "loss": 1.037,
      "step": 10514
    },
    {
      "epoch": 0.7579470914726447,
      "grad_norm": 0.11701200296055876,
      "learning_rate": 2.9190337719057215e-05,
      "loss": 1.0449,
      "step": 10515
    },
    {
      "epoch": 0.7580191739349816,
      "grad_norm": 0.14115412557172433,
      "learning_rate": 2.9173853862592326e-05,
      "loss": 1.0346,
      "step": 10516
    },
    {
      "epoch": 0.7580912563973186,
      "grad_norm": 0.11416853730370119,
      "learning_rate": 2.9157373866786043e-05,
      "loss": 1.0466,
      "step": 10517
    },
    {
      "epoch": 0.7581633388596555,
      "grad_norm": 0.1111710003102329,
      "learning_rate": 2.9140897732536643e-05,
      "loss": 1.0828,
      "step": 10518
    },
    {
      "epoch": 0.7582354213219924,
      "grad_norm": 0.14101327437702532,
      "learning_rate": 2.9124425460742254e-05,
      "loss": 1.047,
      "step": 10519
    },
    {
      "epoch": 0.7583075037843293,
      "grad_norm": 0.1573177294818655,
      "learning_rate": 2.9107957052300783e-05,
      "loss": 1.0369,
      "step": 10520
    },
    {
      "epoch": 0.7583795862466662,
      "grad_norm": 0.15961247165950962,
      "learning_rate": 2.9091492508109795e-05,
      "loss": 0.9977,
      "step": 10521
    },
    {
      "epoch": 0.758451668709003,
      "grad_norm": 0.12053375977869803,
      "learning_rate": 2.9075031829066867e-05,
      "loss": 1.0637,
      "step": 10522
    },
    {
      "epoch": 0.75852375117134,
      "grad_norm": 0.11157912549036648,
      "learning_rate": 2.9058575016069223e-05,
      "loss": 1.0776,
      "step": 10523
    },
    {
      "epoch": 0.7585958336336769,
      "grad_norm": 0.10447324877701583,
      "learning_rate": 2.9042122070013932e-05,
      "loss": 1.0449,
      "step": 10524
    },
    {
      "epoch": 0.7586679160960138,
      "grad_norm": 0.12341068183708785,
      "learning_rate": 2.9025672991797724e-05,
      "loss": 1.0116,
      "step": 10525
    },
    {
      "epoch": 0.7587399985583507,
      "grad_norm": 0.11782392398732332,
      "learning_rate": 2.9009227782317327e-05,
      "loss": 1.0434,
      "step": 10526
    },
    {
      "epoch": 0.7588120810206876,
      "grad_norm": 0.12782609437453255,
      "learning_rate": 2.8992786442469123e-05,
      "loss": 1.0506,
      "step": 10527
    },
    {
      "epoch": 0.7588841634830246,
      "grad_norm": 0.1315482967765847,
      "learning_rate": 2.8976348973149337e-05,
      "loss": 1.0368,
      "step": 10528
    },
    {
      "epoch": 0.7589562459453615,
      "grad_norm": 0.21069124330273237,
      "learning_rate": 2.8959915375253866e-05,
      "loss": 1.1315,
      "step": 10529
    },
    {
      "epoch": 0.7590283284076984,
      "grad_norm": 0.15393115769057733,
      "learning_rate": 2.894348564967857e-05,
      "loss": 1.0521,
      "step": 10530
    },
    {
      "epoch": 0.7591004108700353,
      "grad_norm": 0.19500674277274194,
      "learning_rate": 2.8927059797319002e-05,
      "loss": 1.0318,
      "step": 10531
    },
    {
      "epoch": 0.7591724933323722,
      "grad_norm": 0.23514737965796945,
      "learning_rate": 2.8910637819070552e-05,
      "loss": 0.9678,
      "step": 10532
    },
    {
      "epoch": 0.7592445757947092,
      "grad_norm": 0.12540882957700805,
      "learning_rate": 2.8894219715828286e-05,
      "loss": 1.0613,
      "step": 10533
    },
    {
      "epoch": 0.7593166582570461,
      "grad_norm": 0.11964934923171457,
      "learning_rate": 2.8877805488487143e-05,
      "loss": 0.9972,
      "step": 10534
    },
    {
      "epoch": 0.759388740719383,
      "grad_norm": 0.1972717255294265,
      "learning_rate": 2.8861395137941905e-05,
      "loss": 1.0187,
      "step": 10535
    },
    {
      "epoch": 0.7594608231817199,
      "grad_norm": 0.11360619387148502,
      "learning_rate": 2.8844988665087092e-05,
      "loss": 1.0581,
      "step": 10536
    },
    {
      "epoch": 0.7595329056440568,
      "grad_norm": 0.1350391080317478,
      "learning_rate": 2.8828586070816944e-05,
      "loss": 1.0954,
      "step": 10537
    },
    {
      "epoch": 0.7596049881063938,
      "grad_norm": 0.1688900856447779,
      "learning_rate": 2.881218735602553e-05,
      "loss": 1.0748,
      "step": 10538
    },
    {
      "epoch": 0.7596770705687306,
      "grad_norm": 0.12219526072316197,
      "learning_rate": 2.8795792521606814e-05,
      "loss": 1.0452,
      "step": 10539
    },
    {
      "epoch": 0.7597491530310675,
      "grad_norm": 0.14466854343534055,
      "learning_rate": 2.8779401568454433e-05,
      "loss": 1.1133,
      "step": 10540
    },
    {
      "epoch": 0.7598212354934044,
      "grad_norm": 0.13587312179791383,
      "learning_rate": 2.8763014497461803e-05,
      "loss": 1.1184,
      "step": 10541
    },
    {
      "epoch": 0.7598933179557413,
      "grad_norm": 0.11388666146954698,
      "learning_rate": 2.8746631309522175e-05,
      "loss": 1.035,
      "step": 10542
    },
    {
      "epoch": 0.7599654004180783,
      "grad_norm": 0.11837275288199588,
      "learning_rate": 2.873025200552858e-05,
      "loss": 0.9531,
      "step": 10543
    },
    {
      "epoch": 0.7600374828804152,
      "grad_norm": 0.11424980617306156,
      "learning_rate": 2.87138765863739e-05,
      "loss": 1.0796,
      "step": 10544
    },
    {
      "epoch": 0.7601095653427521,
      "grad_norm": 0.10756183224720135,
      "learning_rate": 2.8697505052950657e-05,
      "loss": 1.0015,
      "step": 10545
    },
    {
      "epoch": 0.760181647805089,
      "grad_norm": 0.1184494532237102,
      "learning_rate": 2.8681137406151282e-05,
      "loss": 1.0616,
      "step": 10546
    },
    {
      "epoch": 0.7602537302674259,
      "grad_norm": 0.15601223048421453,
      "learning_rate": 2.8664773646867915e-05,
      "loss": 1.0305,
      "step": 10547
    },
    {
      "epoch": 0.7603258127297629,
      "grad_norm": 0.13388521882515197,
      "learning_rate": 2.8648413775992644e-05,
      "loss": 1.0372,
      "step": 10548
    },
    {
      "epoch": 0.7603978951920998,
      "grad_norm": 0.11653999227324367,
      "learning_rate": 2.863205779441711e-05,
      "loss": 1.0117,
      "step": 10549
    },
    {
      "epoch": 0.7604699776544367,
      "grad_norm": 0.37519642053291213,
      "learning_rate": 2.8615705703032902e-05,
      "loss": 1.0556,
      "step": 10550
    },
    {
      "epoch": 0.7605420601167736,
      "grad_norm": 0.1259814375367957,
      "learning_rate": 2.859935750273133e-05,
      "loss": 1.0538,
      "step": 10551
    },
    {
      "epoch": 0.7606141425791105,
      "grad_norm": 0.13509925186498598,
      "learning_rate": 2.858301319440361e-05,
      "loss": 1.0158,
      "step": 10552
    },
    {
      "epoch": 0.7606862250414475,
      "grad_norm": 0.1230577412952101,
      "learning_rate": 2.856667277894054e-05,
      "loss": 1.0877,
      "step": 10553
    },
    {
      "epoch": 0.7607583075037844,
      "grad_norm": 0.12800885864326442,
      "learning_rate": 2.855033625723287e-05,
      "loss": 1.0042,
      "step": 10554
    },
    {
      "epoch": 0.7608303899661213,
      "grad_norm": 0.1281989966086143,
      "learning_rate": 2.853400363017108e-05,
      "loss": 1.0426,
      "step": 10555
    },
    {
      "epoch": 0.7609024724284582,
      "grad_norm": 0.12325342935219308,
      "learning_rate": 2.8517674898645452e-05,
      "loss": 1.1228,
      "step": 10556
    },
    {
      "epoch": 0.760974554890795,
      "grad_norm": 0.12767372181589007,
      "learning_rate": 2.850135006354604e-05,
      "loss": 1.0946,
      "step": 10557
    },
    {
      "epoch": 0.7610466373531319,
      "grad_norm": 0.14744625834708455,
      "learning_rate": 2.8485029125762687e-05,
      "loss": 1.1061,
      "step": 10558
    },
    {
      "epoch": 0.7611187198154689,
      "grad_norm": 0.11014184836740534,
      "learning_rate": 2.8468712086185047e-05,
      "loss": 0.9778,
      "step": 10559
    },
    {
      "epoch": 0.7611908022778058,
      "grad_norm": 0.12364072842974796,
      "learning_rate": 2.845239894570253e-05,
      "loss": 1.1026,
      "step": 10560
    },
    {
      "epoch": 0.7612628847401427,
      "grad_norm": 0.12053034936714352,
      "learning_rate": 2.8436089705204348e-05,
      "loss": 1.0381,
      "step": 10561
    },
    {
      "epoch": 0.7613349672024796,
      "grad_norm": 0.12310998920344174,
      "learning_rate": 2.8419784365579516e-05,
      "loss": 1.0507,
      "step": 10562
    },
    {
      "epoch": 0.7614070496648165,
      "grad_norm": 0.20161704533505223,
      "learning_rate": 2.8403482927716807e-05,
      "loss": 0.9888,
      "step": 10563
    },
    {
      "epoch": 0.7614791321271535,
      "grad_norm": 0.12185862573185664,
      "learning_rate": 2.83871853925048e-05,
      "loss": 1.0972,
      "step": 10564
    },
    {
      "epoch": 0.7615512145894904,
      "grad_norm": 0.12948635356799834,
      "learning_rate": 2.8370891760831853e-05,
      "loss": 1.0225,
      "step": 10565
    },
    {
      "epoch": 0.7616232970518273,
      "grad_norm": 0.11441199084811904,
      "learning_rate": 2.835460203358612e-05,
      "loss": 1.0089,
      "step": 10566
    },
    {
      "epoch": 0.7616953795141642,
      "grad_norm": 0.12302033320431575,
      "learning_rate": 2.8338316211655536e-05,
      "loss": 0.9584,
      "step": 10567
    },
    {
      "epoch": 0.7617674619765011,
      "grad_norm": 0.13861045088588828,
      "learning_rate": 2.8322034295927823e-05,
      "loss": 1.0583,
      "step": 10568
    },
    {
      "epoch": 0.7618395444388381,
      "grad_norm": 0.12220781903210876,
      "learning_rate": 2.8305756287290484e-05,
      "loss": 1.0263,
      "step": 10569
    },
    {
      "epoch": 0.761911626901175,
      "grad_norm": 0.1205695373846113,
      "learning_rate": 2.8289482186630823e-05,
      "loss": 1.0998,
      "step": 10570
    },
    {
      "epoch": 0.7619837093635119,
      "grad_norm": 0.12682769640805097,
      "learning_rate": 2.8273211994835923e-05,
      "loss": 1.056,
      "step": 10571
    },
    {
      "epoch": 0.7620557918258488,
      "grad_norm": 0.1458386972400512,
      "learning_rate": 2.825694571279266e-05,
      "loss": 1.0017,
      "step": 10572
    },
    {
      "epoch": 0.7621278742881857,
      "grad_norm": 0.13252132282182186,
      "learning_rate": 2.8240683341387687e-05,
      "loss": 1.0314,
      "step": 10573
    },
    {
      "epoch": 0.7621999567505225,
      "grad_norm": 0.11315853625701781,
      "learning_rate": 2.822442488150745e-05,
      "loss": 1.0816,
      "step": 10574
    },
    {
      "epoch": 0.7622720392128595,
      "grad_norm": 0.130592450585755,
      "learning_rate": 2.8208170334038176e-05,
      "loss": 0.9833,
      "step": 10575
    },
    {
      "epoch": 0.7623441216751964,
      "grad_norm": 0.13757855950850634,
      "learning_rate": 2.8191919699865888e-05,
      "loss": 1.018,
      "step": 10576
    },
    {
      "epoch": 0.7624162041375333,
      "grad_norm": 0.11396205870301662,
      "learning_rate": 2.8175672979876423e-05,
      "loss": 0.9578,
      "step": 10577
    },
    {
      "epoch": 0.7624882865998702,
      "grad_norm": 0.13070369587232053,
      "learning_rate": 2.815943017495528e-05,
      "loss": 1.0009,
      "step": 10578
    },
    {
      "epoch": 0.7625603690622071,
      "grad_norm": 0.1310438143866865,
      "learning_rate": 2.8143191285987926e-05,
      "loss": 0.974,
      "step": 10579
    },
    {
      "epoch": 0.7626324515245441,
      "grad_norm": 0.1575946180790465,
      "learning_rate": 2.8126956313859486e-05,
      "loss": 1.091,
      "step": 10580
    },
    {
      "epoch": 0.762704533986881,
      "grad_norm": 0.12680454514962425,
      "learning_rate": 2.811072525945496e-05,
      "loss": 1.0334,
      "step": 10581
    },
    {
      "epoch": 0.7627766164492179,
      "grad_norm": 0.12235132748729526,
      "learning_rate": 2.8094498123658985e-05,
      "loss": 1.0609,
      "step": 10582
    },
    {
      "epoch": 0.7628486989115548,
      "grad_norm": 0.15042200757369792,
      "learning_rate": 2.8078274907356185e-05,
      "loss": 1.0576,
      "step": 10583
    },
    {
      "epoch": 0.7629207813738917,
      "grad_norm": 0.12213785033991577,
      "learning_rate": 2.8062055611430825e-05,
      "loss": 0.9568,
      "step": 10584
    },
    {
      "epoch": 0.7629928638362287,
      "grad_norm": 0.1312087827802585,
      "learning_rate": 2.804584023676704e-05,
      "loss": 0.969,
      "step": 10585
    },
    {
      "epoch": 0.7630649462985656,
      "grad_norm": 0.11976410201329941,
      "learning_rate": 2.8029628784248653e-05,
      "loss": 1.0252,
      "step": 10586
    },
    {
      "epoch": 0.7631370287609025,
      "grad_norm": 0.12369048344896541,
      "learning_rate": 2.801342125475933e-05,
      "loss": 1.0241,
      "step": 10587
    },
    {
      "epoch": 0.7632091112232394,
      "grad_norm": 0.12185463096888792,
      "learning_rate": 2.799721764918263e-05,
      "loss": 1.0059,
      "step": 10588
    },
    {
      "epoch": 0.7632811936855763,
      "grad_norm": 0.1085591347016883,
      "learning_rate": 2.7981017968401678e-05,
      "loss": 0.9985,
      "step": 10589
    },
    {
      "epoch": 0.7633532761479133,
      "grad_norm": 0.13176137711225747,
      "learning_rate": 2.7964822213299556e-05,
      "loss": 1.0285,
      "step": 10590
    },
    {
      "epoch": 0.7634253586102501,
      "grad_norm": 0.12441706348393314,
      "learning_rate": 2.7948630384759035e-05,
      "loss": 1.0667,
      "step": 10591
    },
    {
      "epoch": 0.763497441072587,
      "grad_norm": 0.12439120572232766,
      "learning_rate": 2.793244248366281e-05,
      "loss": 1.0569,
      "step": 10592
    },
    {
      "epoch": 0.7635695235349239,
      "grad_norm": 0.13733245426286628,
      "learning_rate": 2.7916258510893168e-05,
      "loss": 1.0314,
      "step": 10593
    },
    {
      "epoch": 0.7636416059972608,
      "grad_norm": 0.13669743786806246,
      "learning_rate": 2.7900078467332323e-05,
      "loss": 1.0215,
      "step": 10594
    },
    {
      "epoch": 0.7637136884595978,
      "grad_norm": 0.16503022985131632,
      "learning_rate": 2.78839023538622e-05,
      "loss": 1.1017,
      "step": 10595
    },
    {
      "epoch": 0.7637857709219347,
      "grad_norm": 0.12335908174946014,
      "learning_rate": 2.7867730171364624e-05,
      "loss": 0.9892,
      "step": 10596
    },
    {
      "epoch": 0.7638578533842716,
      "grad_norm": 0.11800150991126343,
      "learning_rate": 2.7851561920721035e-05,
      "loss": 1.0729,
      "step": 10597
    },
    {
      "epoch": 0.7639299358466085,
      "grad_norm": 0.13883812296962822,
      "learning_rate": 2.7835397602812786e-05,
      "loss": 1.0852,
      "step": 10598
    },
    {
      "epoch": 0.7640020183089454,
      "grad_norm": 0.11890452400758561,
      "learning_rate": 2.7819237218520966e-05,
      "loss": 0.9578,
      "step": 10599
    },
    {
      "epoch": 0.7640741007712823,
      "grad_norm": 0.15751024181839374,
      "learning_rate": 2.7803080768726462e-05,
      "loss": 0.9357,
      "step": 10600
    },
    {
      "epoch": 0.7641461832336193,
      "grad_norm": 0.12965455423226463,
      "learning_rate": 2.7786928254309953e-05,
      "loss": 0.9705,
      "step": 10601
    },
    {
      "epoch": 0.7642182656959562,
      "grad_norm": 0.13327413168072563,
      "learning_rate": 2.777077967615188e-05,
      "loss": 0.9758,
      "step": 10602
    },
    {
      "epoch": 0.7642903481582931,
      "grad_norm": 0.1117838188918969,
      "learning_rate": 2.7754635035132514e-05,
      "loss": 0.8837,
      "step": 10603
    },
    {
      "epoch": 0.76436243062063,
      "grad_norm": 0.11780624768581949,
      "learning_rate": 2.7738494332131846e-05,
      "loss": 1.0625,
      "step": 10604
    },
    {
      "epoch": 0.764434513082967,
      "grad_norm": 0.27469513568968457,
      "learning_rate": 2.7722357568029723e-05,
      "loss": 1.0535,
      "step": 10605
    },
    {
      "epoch": 0.7645065955453039,
      "grad_norm": 0.13200393846775857,
      "learning_rate": 2.7706224743705724e-05,
      "loss": 1.0139,
      "step": 10606
    },
    {
      "epoch": 0.7645786780076408,
      "grad_norm": 0.11943778274304587,
      "learning_rate": 2.7690095860039224e-05,
      "loss": 1.0896,
      "step": 10607
    },
    {
      "epoch": 0.7646507604699777,
      "grad_norm": 0.13147865262057573,
      "learning_rate": 2.767397091790941e-05,
      "loss": 1.1116,
      "step": 10608
    },
    {
      "epoch": 0.7647228429323145,
      "grad_norm": 0.12024942305855452,
      "learning_rate": 2.7657849918195223e-05,
      "loss": 1.0274,
      "step": 10609
    },
    {
      "epoch": 0.7647949253946514,
      "grad_norm": 0.11344776450943779,
      "learning_rate": 2.7641732861775414e-05,
      "loss": 0.9864,
      "step": 10610
    },
    {
      "epoch": 0.7648670078569884,
      "grad_norm": 0.139185568575025,
      "learning_rate": 2.76256197495285e-05,
      "loss": 1.0243,
      "step": 10611
    },
    {
      "epoch": 0.7649390903193253,
      "grad_norm": 0.1256239457917406,
      "learning_rate": 2.7609510582332788e-05,
      "loss": 1.0179,
      "step": 10612
    },
    {
      "epoch": 0.7650111727816622,
      "grad_norm": 0.12079551723573129,
      "learning_rate": 2.759340536106637e-05,
      "loss": 1.0448,
      "step": 10613
    },
    {
      "epoch": 0.7650832552439991,
      "grad_norm": 0.11649357608715397,
      "learning_rate": 2.7577304086607125e-05,
      "loss": 1.0283,
      "step": 10614
    },
    {
      "epoch": 0.765155337706336,
      "grad_norm": 0.12992743177301191,
      "learning_rate": 2.7561206759832724e-05,
      "loss": 1.0891,
      "step": 10615
    },
    {
      "epoch": 0.765227420168673,
      "grad_norm": 0.13940112427214507,
      "learning_rate": 2.7545113381620612e-05,
      "loss": 0.993,
      "step": 10616
    },
    {
      "epoch": 0.7652995026310099,
      "grad_norm": 0.21484730511146136,
      "learning_rate": 2.752902395284801e-05,
      "loss": 1.0846,
      "step": 10617
    },
    {
      "epoch": 0.7653715850933468,
      "grad_norm": 0.1269998989153757,
      "learning_rate": 2.751293847439196e-05,
      "loss": 1.0835,
      "step": 10618
    },
    {
      "epoch": 0.7654436675556837,
      "grad_norm": 0.12054861372266659,
      "learning_rate": 2.7496856947129247e-05,
      "loss": 1.0399,
      "step": 10619
    },
    {
      "epoch": 0.7655157500180206,
      "grad_norm": 0.11778560276209397,
      "learning_rate": 2.7480779371936462e-05,
      "loss": 1.0413,
      "step": 10620
    },
    {
      "epoch": 0.7655878324803576,
      "grad_norm": 0.12668449368212503,
      "learning_rate": 2.7464705749690002e-05,
      "loss": 0.9901,
      "step": 10621
    },
    {
      "epoch": 0.7656599149426945,
      "grad_norm": 0.11952043064910972,
      "learning_rate": 2.7448636081265942e-05,
      "loss": 1.0185,
      "step": 10622
    },
    {
      "epoch": 0.7657319974050314,
      "grad_norm": 0.10352091726979745,
      "learning_rate": 2.743257036754031e-05,
      "loss": 1.0349,
      "step": 10623
    },
    {
      "epoch": 0.7658040798673683,
      "grad_norm": 0.12128838640807624,
      "learning_rate": 2.741650860938879e-05,
      "loss": 0.9655,
      "step": 10624
    },
    {
      "epoch": 0.7658761623297052,
      "grad_norm": 0.14225285501192259,
      "learning_rate": 2.7400450807686938e-05,
      "loss": 1.0472,
      "step": 10625
    },
    {
      "epoch": 0.765948244792042,
      "grad_norm": 0.13155370622341392,
      "learning_rate": 2.7384396963309955e-05,
      "loss": 0.9357,
      "step": 10626
    },
    {
      "epoch": 0.766020327254379,
      "grad_norm": 0.11387106469225848,
      "learning_rate": 2.7368347077133004e-05,
      "loss": 0.9731,
      "step": 10627
    },
    {
      "epoch": 0.7660924097167159,
      "grad_norm": 0.12326112840726766,
      "learning_rate": 2.7352301150030912e-05,
      "loss": 1.1081,
      "step": 10628
    },
    {
      "epoch": 0.7661644921790528,
      "grad_norm": 0.1299285675955688,
      "learning_rate": 2.7336259182878376e-05,
      "loss": 1.0672,
      "step": 10629
    },
    {
      "epoch": 0.7662365746413897,
      "grad_norm": 0.11690576472807482,
      "learning_rate": 2.7320221176549744e-05,
      "loss": 1.0509,
      "step": 10630
    },
    {
      "epoch": 0.7663086571037266,
      "grad_norm": 0.507374063012129,
      "learning_rate": 2.7304187131919235e-05,
      "loss": 1.0893,
      "step": 10631
    },
    {
      "epoch": 0.7663807395660636,
      "grad_norm": 0.1692676117210617,
      "learning_rate": 2.728815704986092e-05,
      "loss": 1.15,
      "step": 10632
    },
    {
      "epoch": 0.7664528220284005,
      "grad_norm": 0.11339256355629325,
      "learning_rate": 2.7272130931248573e-05,
      "loss": 1.0036,
      "step": 10633
    },
    {
      "epoch": 0.7665249044907374,
      "grad_norm": 0.1296208498183333,
      "learning_rate": 2.7256108776955702e-05,
      "loss": 1.0208,
      "step": 10634
    },
    {
      "epoch": 0.7665969869530743,
      "grad_norm": 0.13762937267204073,
      "learning_rate": 2.7240090587855647e-05,
      "loss": 1.0351,
      "step": 10635
    },
    {
      "epoch": 0.7666690694154112,
      "grad_norm": 0.12368988769575019,
      "learning_rate": 2.7224076364821627e-05,
      "loss": 1.0942,
      "step": 10636
    },
    {
      "epoch": 0.7667411518777482,
      "grad_norm": 0.12145923433424148,
      "learning_rate": 2.720806610872655e-05,
      "loss": 1.0553,
      "step": 10637
    },
    {
      "epoch": 0.7668132343400851,
      "grad_norm": 0.1194039343710927,
      "learning_rate": 2.719205982044305e-05,
      "loss": 1.0209,
      "step": 10638
    },
    {
      "epoch": 0.766885316802422,
      "grad_norm": 0.14668953156745324,
      "learning_rate": 2.717605750084361e-05,
      "loss": 1.0911,
      "step": 10639
    },
    {
      "epoch": 0.7669573992647589,
      "grad_norm": 0.13450832853239833,
      "learning_rate": 2.7160059150800578e-05,
      "loss": 0.9831,
      "step": 10640
    },
    {
      "epoch": 0.7670294817270958,
      "grad_norm": 0.16633627789033437,
      "learning_rate": 2.7144064771185997e-05,
      "loss": 1.0616,
      "step": 10641
    },
    {
      "epoch": 0.7671015641894328,
      "grad_norm": 0.11553029349439153,
      "learning_rate": 2.7128074362871646e-05,
      "loss": 0.9951,
      "step": 10642
    },
    {
      "epoch": 0.7671736466517696,
      "grad_norm": 0.2680187478120073,
      "learning_rate": 2.7112087926729168e-05,
      "loss": 1.0867,
      "step": 10643
    },
    {
      "epoch": 0.7672457291141065,
      "grad_norm": 0.12462742486886384,
      "learning_rate": 2.7096105463629938e-05,
      "loss": 1.0733,
      "step": 10644
    },
    {
      "epoch": 0.7673178115764434,
      "grad_norm": 0.16283651556219533,
      "learning_rate": 2.7080126974445242e-05,
      "loss": 0.9314,
      "step": 10645
    },
    {
      "epoch": 0.7673898940387803,
      "grad_norm": 0.11441700107824704,
      "learning_rate": 2.7064152460045955e-05,
      "loss": 0.9815,
      "step": 10646
    },
    {
      "epoch": 0.7674619765011172,
      "grad_norm": 0.13601833159141044,
      "learning_rate": 2.704818192130286e-05,
      "loss": 1.0793,
      "step": 10647
    },
    {
      "epoch": 0.7675340589634542,
      "grad_norm": 0.12269644252993181,
      "learning_rate": 2.7032215359086465e-05,
      "loss": 1.0562,
      "step": 10648
    },
    {
      "epoch": 0.7676061414257911,
      "grad_norm": 0.14626904175207678,
      "learning_rate": 2.7016252774267192e-05,
      "loss": 1.0006,
      "step": 10649
    },
    {
      "epoch": 0.767678223888128,
      "grad_norm": 0.14693118156012258,
      "learning_rate": 2.7000294167715034e-05,
      "loss": 1.0256,
      "step": 10650
    },
    {
      "epoch": 0.7677503063504649,
      "grad_norm": 0.118477437805503,
      "learning_rate": 2.698433954029992e-05,
      "loss": 1.0672,
      "step": 10651
    },
    {
      "epoch": 0.7678223888128018,
      "grad_norm": 0.5407071926983716,
      "learning_rate": 2.6968388892891516e-05,
      "loss": 1.0304,
      "step": 10652
    },
    {
      "epoch": 0.7678944712751388,
      "grad_norm": 0.13239574411320817,
      "learning_rate": 2.6952442226359277e-05,
      "loss": 0.9912,
      "step": 10653
    },
    {
      "epoch": 0.7679665537374757,
      "grad_norm": 0.1263685288270311,
      "learning_rate": 2.6936499541572445e-05,
      "loss": 1.0074,
      "step": 10654
    },
    {
      "epoch": 0.7680386361998126,
      "grad_norm": 0.10128075670000627,
      "learning_rate": 2.6920560839400023e-05,
      "loss": 0.9556,
      "step": 10655
    },
    {
      "epoch": 0.7681107186621495,
      "grad_norm": 0.1303988970197323,
      "learning_rate": 2.690462612071083e-05,
      "loss": 1.1159,
      "step": 10656
    },
    {
      "epoch": 0.7681828011244864,
      "grad_norm": 0.12657033752468821,
      "learning_rate": 2.688869538637344e-05,
      "loss": 1.0572,
      "step": 10657
    },
    {
      "epoch": 0.7682548835868234,
      "grad_norm": 0.12851288811210584,
      "learning_rate": 2.687276863725623e-05,
      "loss": 1.0115,
      "step": 10658
    },
    {
      "epoch": 0.7683269660491603,
      "grad_norm": 0.1492492721242618,
      "learning_rate": 2.6856845874227333e-05,
      "loss": 1.0709,
      "step": 10659
    },
    {
      "epoch": 0.7683990485114972,
      "grad_norm": 0.1296877233216076,
      "learning_rate": 2.68409270981547e-05,
      "loss": 1.0102,
      "step": 10660
    },
    {
      "epoch": 0.768471130973834,
      "grad_norm": 0.11417989746257491,
      "learning_rate": 2.682501230990604e-05,
      "loss": 1.0447,
      "step": 10661
    },
    {
      "epoch": 0.7685432134361709,
      "grad_norm": 0.14728763378226734,
      "learning_rate": 2.6809101510348857e-05,
      "loss": 1.0374,
      "step": 10662
    },
    {
      "epoch": 0.7686152958985079,
      "grad_norm": 0.17765517223258626,
      "learning_rate": 2.6793194700350422e-05,
      "loss": 1.0501,
      "step": 10663
    },
    {
      "epoch": 0.7686873783608448,
      "grad_norm": 0.1085226897118502,
      "learning_rate": 2.67772918807778e-05,
      "loss": 0.9557,
      "step": 10664
    },
    {
      "epoch": 0.7687594608231817,
      "grad_norm": 0.11240492787668165,
      "learning_rate": 2.676139305249783e-05,
      "loss": 1.0923,
      "step": 10665
    },
    {
      "epoch": 0.7688315432855186,
      "grad_norm": 0.2883310472576743,
      "learning_rate": 2.6745498216377163e-05,
      "loss": 1.1147,
      "step": 10666
    },
    {
      "epoch": 0.7689036257478555,
      "grad_norm": 0.13380674367221126,
      "learning_rate": 2.6729607373282193e-05,
      "loss": 1.1053,
      "step": 10667
    },
    {
      "epoch": 0.7689757082101925,
      "grad_norm": 0.10471823974673071,
      "learning_rate": 2.671372052407912e-05,
      "loss": 1.0896,
      "step": 10668
    },
    {
      "epoch": 0.7690477906725294,
      "grad_norm": 0.14726400970922088,
      "learning_rate": 2.669783766963392e-05,
      "loss": 1.0344,
      "step": 10669
    },
    {
      "epoch": 0.7691198731348663,
      "grad_norm": 0.11685826260412144,
      "learning_rate": 2.6681958810812336e-05,
      "loss": 1.1086,
      "step": 10670
    },
    {
      "epoch": 0.7691919555972032,
      "grad_norm": 0.10714857798514922,
      "learning_rate": 2.6666083948479924e-05,
      "loss": 1.0538,
      "step": 10671
    },
    {
      "epoch": 0.7692640380595401,
      "grad_norm": 0.13727574463817657,
      "learning_rate": 2.6650213083501995e-05,
      "loss": 1.0082,
      "step": 10672
    },
    {
      "epoch": 0.769336120521877,
      "grad_norm": 0.12732226445424788,
      "learning_rate": 2.6634346216743665e-05,
      "loss": 1.0114,
      "step": 10673
    },
    {
      "epoch": 0.769408202984214,
      "grad_norm": 0.13951048972232055,
      "learning_rate": 2.6618483349069846e-05,
      "loss": 1.0963,
      "step": 10674
    },
    {
      "epoch": 0.7694802854465509,
      "grad_norm": 0.1293788784265846,
      "learning_rate": 2.6602624481345108e-05,
      "loss": 1.0148,
      "step": 10675
    },
    {
      "epoch": 0.7695523679088878,
      "grad_norm": 0.11451204441932056,
      "learning_rate": 2.6586769614434003e-05,
      "loss": 1.0447,
      "step": 10676
    },
    {
      "epoch": 0.7696244503712247,
      "grad_norm": 0.1208818451893575,
      "learning_rate": 2.6570918749200722e-05,
      "loss": 1.0695,
      "step": 10677
    },
    {
      "epoch": 0.7696965328335615,
      "grad_norm": 0.18453005605297662,
      "learning_rate": 2.655507188650932e-05,
      "loss": 1.002,
      "step": 10678
    },
    {
      "epoch": 0.7697686152958985,
      "grad_norm": 0.1497392719239171,
      "learning_rate": 2.65392290272235e-05,
      "loss": 1.0479,
      "step": 10679
    },
    {
      "epoch": 0.7698406977582354,
      "grad_norm": 0.14286175452778913,
      "learning_rate": 2.6523390172206918e-05,
      "loss": 1.0244,
      "step": 10680
    },
    {
      "epoch": 0.7699127802205723,
      "grad_norm": 0.1294040956981068,
      "learning_rate": 2.6507555322322918e-05,
      "loss": 1.0111,
      "step": 10681
    },
    {
      "epoch": 0.7699848626829092,
      "grad_norm": 0.12821440271845203,
      "learning_rate": 2.6491724478434666e-05,
      "loss": 1.0628,
      "step": 10682
    },
    {
      "epoch": 0.7700569451452461,
      "grad_norm": 0.1137834690884793,
      "learning_rate": 2.647589764140499e-05,
      "loss": 1.0014,
      "step": 10683
    },
    {
      "epoch": 0.7701290276075831,
      "grad_norm": 0.1266906062469809,
      "learning_rate": 2.646007481209668e-05,
      "loss": 1.0437,
      "step": 10684
    },
    {
      "epoch": 0.77020111006992,
      "grad_norm": 0.11677186081738636,
      "learning_rate": 2.6444255991372214e-05,
      "loss": 1.0236,
      "step": 10685
    },
    {
      "epoch": 0.7702731925322569,
      "grad_norm": 0.1306532189142619,
      "learning_rate": 2.6428441180093866e-05,
      "loss": 1.0768,
      "step": 10686
    },
    {
      "epoch": 0.7703452749945938,
      "grad_norm": 0.11613380584032591,
      "learning_rate": 2.6412630379123637e-05,
      "loss": 0.9879,
      "step": 10687
    },
    {
      "epoch": 0.7704173574569307,
      "grad_norm": 0.16253373920142306,
      "learning_rate": 2.639682358932336e-05,
      "loss": 1.136,
      "step": 10688
    },
    {
      "epoch": 0.7704894399192677,
      "grad_norm": 0.12192588192735819,
      "learning_rate": 2.638102081155469e-05,
      "loss": 1.0963,
      "step": 10689
    },
    {
      "epoch": 0.7705615223816046,
      "grad_norm": 0.3201214939589248,
      "learning_rate": 2.636522204667904e-05,
      "loss": 1.0229,
      "step": 10690
    },
    {
      "epoch": 0.7706336048439415,
      "grad_norm": 0.10746667023479531,
      "learning_rate": 2.6349427295557506e-05,
      "loss": 1.0689,
      "step": 10691
    },
    {
      "epoch": 0.7707056873062784,
      "grad_norm": 0.12778978830283233,
      "learning_rate": 2.6333636559051046e-05,
      "loss": 1.0206,
      "step": 10692
    },
    {
      "epoch": 0.7707777697686153,
      "grad_norm": 0.11571533255471633,
      "learning_rate": 2.631784983802047e-05,
      "loss": 0.9704,
      "step": 10693
    },
    {
      "epoch": 0.7708498522309523,
      "grad_norm": 0.13079282803930972,
      "learning_rate": 2.6302067133326293e-05,
      "loss": 1.0243,
      "step": 10694
    },
    {
      "epoch": 0.7709219346932891,
      "grad_norm": 0.12115848536666998,
      "learning_rate": 2.6286288445828734e-05,
      "loss": 0.9652,
      "step": 10695
    },
    {
      "epoch": 0.770994017155626,
      "grad_norm": 0.15225295452710813,
      "learning_rate": 2.6270513776387927e-05,
      "loss": 0.9493,
      "step": 10696
    },
    {
      "epoch": 0.7710660996179629,
      "grad_norm": 0.12054325209448234,
      "learning_rate": 2.625474312586369e-05,
      "loss": 1.0839,
      "step": 10697
    },
    {
      "epoch": 0.7711381820802998,
      "grad_norm": 0.11750116567865282,
      "learning_rate": 2.6238976495115754e-05,
      "loss": 1.0506,
      "step": 10698
    },
    {
      "epoch": 0.7712102645426367,
      "grad_norm": 0.11274049945285776,
      "learning_rate": 2.622321388500345e-05,
      "loss": 0.9971,
      "step": 10699
    },
    {
      "epoch": 0.7712823470049737,
      "grad_norm": 0.11823874122788765,
      "learning_rate": 2.620745529638603e-05,
      "loss": 1.0842,
      "step": 10700
    },
    {
      "epoch": 0.7713544294673106,
      "grad_norm": 0.1175799084818747,
      "learning_rate": 2.6191700730122416e-05,
      "loss": 1.0519,
      "step": 10701
    },
    {
      "epoch": 0.7714265119296475,
      "grad_norm": 0.24426370866210856,
      "learning_rate": 2.6175950187071487e-05,
      "loss": 1.0541,
      "step": 10702
    },
    {
      "epoch": 0.7714985943919844,
      "grad_norm": 0.13852102807761202,
      "learning_rate": 2.61602036680917e-05,
      "loss": 1.0048,
      "step": 10703
    },
    {
      "epoch": 0.7715706768543213,
      "grad_norm": 0.1141022150825231,
      "learning_rate": 2.614446117404139e-05,
      "loss": 1.0204,
      "step": 10704
    },
    {
      "epoch": 0.7716427593166583,
      "grad_norm": 0.14291040669795027,
      "learning_rate": 2.6128722705778653e-05,
      "loss": 1.0703,
      "step": 10705
    },
    {
      "epoch": 0.7717148417789952,
      "grad_norm": 0.24493750166335743,
      "learning_rate": 2.6112988264161454e-05,
      "loss": 0.9507,
      "step": 10706
    },
    {
      "epoch": 0.7717869242413321,
      "grad_norm": 0.18899026662331256,
      "learning_rate": 2.6097257850047386e-05,
      "loss": 1.0132,
      "step": 10707
    },
    {
      "epoch": 0.771859006703669,
      "grad_norm": 0.1272933666908047,
      "learning_rate": 2.6081531464293906e-05,
      "loss": 1.0324,
      "step": 10708
    },
    {
      "epoch": 0.771931089166006,
      "grad_norm": 0.11169476382342026,
      "learning_rate": 2.6065809107758266e-05,
      "loss": 1.0115,
      "step": 10709
    },
    {
      "epoch": 0.7720031716283429,
      "grad_norm": 0.11673750109976265,
      "learning_rate": 2.6050090781297464e-05,
      "loss": 1.0233,
      "step": 10710
    },
    {
      "epoch": 0.7720752540906798,
      "grad_norm": 0.17476339069874805,
      "learning_rate": 2.6034376485768286e-05,
      "loss": 1.0518,
      "step": 10711
    },
    {
      "epoch": 0.7721473365530167,
      "grad_norm": 0.12650920550182332,
      "learning_rate": 2.60186662220273e-05,
      "loss": 1.0139,
      "step": 10712
    },
    {
      "epoch": 0.7722194190153535,
      "grad_norm": 0.12742490122856964,
      "learning_rate": 2.6002959990930864e-05,
      "loss": 1.0701,
      "step": 10713
    },
    {
      "epoch": 0.7722915014776904,
      "grad_norm": 0.1610767803582473,
      "learning_rate": 2.5987257793335107e-05,
      "loss": 1.087,
      "step": 10714
    },
    {
      "epoch": 0.7723635839400274,
      "grad_norm": 0.11136058868624245,
      "learning_rate": 2.5971559630095932e-05,
      "loss": 1.0363,
      "step": 10715
    },
    {
      "epoch": 0.7724356664023643,
      "grad_norm": 0.11385182035376569,
      "learning_rate": 2.5955865502069043e-05,
      "loss": 1.0206,
      "step": 10716
    },
    {
      "epoch": 0.7725077488647012,
      "grad_norm": 0.1278533957683299,
      "learning_rate": 2.59401754101099e-05,
      "loss": 1.0146,
      "step": 10717
    },
    {
      "epoch": 0.7725798313270381,
      "grad_norm": 0.14123658794136873,
      "learning_rate": 2.5924489355073744e-05,
      "loss": 0.9554,
      "step": 10718
    },
    {
      "epoch": 0.772651913789375,
      "grad_norm": 0.13959683467518644,
      "learning_rate": 2.5908807337815632e-05,
      "loss": 1.0456,
      "step": 10719
    },
    {
      "epoch": 0.772723996251712,
      "grad_norm": 0.11626771607679076,
      "learning_rate": 2.589312935919035e-05,
      "loss": 0.9501,
      "step": 10720
    },
    {
      "epoch": 0.7727960787140489,
      "grad_norm": 0.1282981047695549,
      "learning_rate": 2.58774554200525e-05,
      "loss": 1.0525,
      "step": 10721
    },
    {
      "epoch": 0.7728681611763858,
      "grad_norm": 0.18042114300152667,
      "learning_rate": 2.586178552125644e-05,
      "loss": 1.0217,
      "step": 10722
    },
    {
      "epoch": 0.7729402436387227,
      "grad_norm": 0.11898500131788246,
      "learning_rate": 2.5846119663656343e-05,
      "loss": 1.0692,
      "step": 10723
    },
    {
      "epoch": 0.7730123261010596,
      "grad_norm": 0.1173487784682748,
      "learning_rate": 2.5830457848106116e-05,
      "loss": 1.0826,
      "step": 10724
    },
    {
      "epoch": 0.7730844085633966,
      "grad_norm": 0.14842993933002066,
      "learning_rate": 2.5814800075459477e-05,
      "loss": 0.9924,
      "step": 10725
    },
    {
      "epoch": 0.7731564910257335,
      "grad_norm": 0.12781665467589015,
      "learning_rate": 2.5799146346569914e-05,
      "loss": 1.018,
      "step": 10726
    },
    {
      "epoch": 0.7732285734880704,
      "grad_norm": 0.12954294263661986,
      "learning_rate": 2.5783496662290695e-05,
      "loss": 1.0701,
      "step": 10727
    },
    {
      "epoch": 0.7733006559504073,
      "grad_norm": 0.13511899548132977,
      "learning_rate": 2.5767851023474865e-05,
      "loss": 1.0089,
      "step": 10728
    },
    {
      "epoch": 0.7733727384127442,
      "grad_norm": 0.14939877430045267,
      "learning_rate": 2.575220943097526e-05,
      "loss": 1.0897,
      "step": 10729
    },
    {
      "epoch": 0.773444820875081,
      "grad_norm": 0.1546589654114307,
      "learning_rate": 2.5736571885644477e-05,
      "loss": 0.9898,
      "step": 10730
    },
    {
      "epoch": 0.773516903337418,
      "grad_norm": 0.10799039287210445,
      "learning_rate": 2.5720938388334937e-05,
      "loss": 1.1593,
      "step": 10731
    },
    {
      "epoch": 0.7735889857997549,
      "grad_norm": 0.1187651669208129,
      "learning_rate": 2.570530893989871e-05,
      "loss": 1.0636,
      "step": 10732
    },
    {
      "epoch": 0.7736610682620918,
      "grad_norm": 0.12591427221637086,
      "learning_rate": 2.5689683541187838e-05,
      "loss": 1.0021,
      "step": 10733
    },
    {
      "epoch": 0.7737331507244287,
      "grad_norm": 0.12941022843223723,
      "learning_rate": 2.5674062193054016e-05,
      "loss": 1.0313,
      "step": 10734
    },
    {
      "epoch": 0.7738052331867656,
      "grad_norm": 0.13699375693752236,
      "learning_rate": 2.5658444896348766e-05,
      "loss": 0.9953,
      "step": 10735
    },
    {
      "epoch": 0.7738773156491026,
      "grad_norm": 0.2618851940311601,
      "learning_rate": 2.5642831651923295e-05,
      "loss": 1.0527,
      "step": 10736
    },
    {
      "epoch": 0.7739493981114395,
      "grad_norm": 0.12968104915129786,
      "learning_rate": 2.5627222460628742e-05,
      "loss": 1.0368,
      "step": 10737
    },
    {
      "epoch": 0.7740214805737764,
      "grad_norm": 0.15004868103437294,
      "learning_rate": 2.561161732331593e-05,
      "loss": 1.0399,
      "step": 10738
    },
    {
      "epoch": 0.7740935630361133,
      "grad_norm": 0.1527497897842376,
      "learning_rate": 2.5596016240835486e-05,
      "loss": 1.1016,
      "step": 10739
    },
    {
      "epoch": 0.7741656454984502,
      "grad_norm": 0.15828360695975174,
      "learning_rate": 2.5580419214037777e-05,
      "loss": 0.953,
      "step": 10740
    },
    {
      "epoch": 0.7742377279607872,
      "grad_norm": 0.13482732634858843,
      "learning_rate": 2.5564826243772966e-05,
      "loss": 1.0298,
      "step": 10741
    },
    {
      "epoch": 0.7743098104231241,
      "grad_norm": 0.15516630449591773,
      "learning_rate": 2.5549237330891097e-05,
      "loss": 1.0283,
      "step": 10742
    },
    {
      "epoch": 0.774381892885461,
      "grad_norm": 0.14394425749215398,
      "learning_rate": 2.553365247624182e-05,
      "loss": 1.0784,
      "step": 10743
    },
    {
      "epoch": 0.7744539753477979,
      "grad_norm": 0.15451622457297784,
      "learning_rate": 2.5518071680674683e-05,
      "loss": 1.0369,
      "step": 10744
    },
    {
      "epoch": 0.7745260578101348,
      "grad_norm": 0.12014612893986848,
      "learning_rate": 2.5502494945038934e-05,
      "loss": 1.0501,
      "step": 10745
    },
    {
      "epoch": 0.7745981402724718,
      "grad_norm": 0.15741023650205782,
      "learning_rate": 2.5486922270183756e-05,
      "loss": 1.0495,
      "step": 10746
    },
    {
      "epoch": 0.7746702227348086,
      "grad_norm": 0.12218256403561806,
      "learning_rate": 2.5471353656957886e-05,
      "loss": 1.0057,
      "step": 10747
    },
    {
      "epoch": 0.7747423051971455,
      "grad_norm": 0.15025879580787882,
      "learning_rate": 2.5455789106210004e-05,
      "loss": 1.041,
      "step": 10748
    },
    {
      "epoch": 0.7748143876594824,
      "grad_norm": 0.12397832302151543,
      "learning_rate": 2.5440228618788465e-05,
      "loss": 1.0055,
      "step": 10749
    },
    {
      "epoch": 0.7748864701218193,
      "grad_norm": 0.1359455015142996,
      "learning_rate": 2.5424672195541565e-05,
      "loss": 1.0718,
      "step": 10750
    },
    {
      "epoch": 0.7749585525841562,
      "grad_norm": 0.11967710146309664,
      "learning_rate": 2.5409119837317173e-05,
      "loss": 1.0844,
      "step": 10751
    },
    {
      "epoch": 0.7750306350464932,
      "grad_norm": 0.13551162808618972,
      "learning_rate": 2.5393571544963056e-05,
      "loss": 0.9916,
      "step": 10752
    },
    {
      "epoch": 0.7751027175088301,
      "grad_norm": 0.2505113131520362,
      "learning_rate": 2.5378027319326736e-05,
      "loss": 0.9954,
      "step": 10753
    },
    {
      "epoch": 0.775174799971167,
      "grad_norm": 0.13010339564352216,
      "learning_rate": 2.536248716125552e-05,
      "loss": 1.1229,
      "step": 10754
    },
    {
      "epoch": 0.7752468824335039,
      "grad_norm": 0.12078507783000458,
      "learning_rate": 2.534695107159647e-05,
      "loss": 1.0791,
      "step": 10755
    },
    {
      "epoch": 0.7753189648958408,
      "grad_norm": 0.15261616683023113,
      "learning_rate": 2.5331419051196458e-05,
      "loss": 1.0489,
      "step": 10756
    },
    {
      "epoch": 0.7753910473581778,
      "grad_norm": 0.1438182650983518,
      "learning_rate": 2.531589110090211e-05,
      "loss": 0.9755,
      "step": 10757
    },
    {
      "epoch": 0.7754631298205147,
      "grad_norm": 0.10770321471879783,
      "learning_rate": 2.530036722155985e-05,
      "loss": 1.0531,
      "step": 10758
    },
    {
      "epoch": 0.7755352122828516,
      "grad_norm": 0.3004573127195272,
      "learning_rate": 2.5284847414015845e-05,
      "loss": 1.0857,
      "step": 10759
    },
    {
      "epoch": 0.7756072947451885,
      "grad_norm": 0.12583486482848552,
      "learning_rate": 2.5269331679116082e-05,
      "loss": 0.999,
      "step": 10760
    },
    {
      "epoch": 0.7756793772075254,
      "grad_norm": 0.12081470404250179,
      "learning_rate": 2.525382001770631e-05,
      "loss": 1.0205,
      "step": 10761
    },
    {
      "epoch": 0.7757514596698624,
      "grad_norm": 0.10826130831984797,
      "learning_rate": 2.523831243063204e-05,
      "loss": 0.9961,
      "step": 10762
    },
    {
      "epoch": 0.7758235421321993,
      "grad_norm": 0.12722705786452104,
      "learning_rate": 2.5222808918738573e-05,
      "loss": 0.9849,
      "step": 10763
    },
    {
      "epoch": 0.7758956245945362,
      "grad_norm": 0.23862529381510056,
      "learning_rate": 2.5207309482871e-05,
      "loss": 1.0171,
      "step": 10764
    },
    {
      "epoch": 0.775967707056873,
      "grad_norm": 0.1151429312852684,
      "learning_rate": 2.519181412387418e-05,
      "loss": 1.0341,
      "step": 10765
    },
    {
      "epoch": 0.7760397895192099,
      "grad_norm": 0.14299972834226815,
      "learning_rate": 2.5176322842592738e-05,
      "loss": 1.1283,
      "step": 10766
    },
    {
      "epoch": 0.7761118719815469,
      "grad_norm": 0.1315349362261664,
      "learning_rate": 2.5160835639871084e-05,
      "loss": 1.0553,
      "step": 10767
    },
    {
      "epoch": 0.7761839544438838,
      "grad_norm": 0.1406557735809674,
      "learning_rate": 2.5145352516553434e-05,
      "loss": 1.0306,
      "step": 10768
    },
    {
      "epoch": 0.7762560369062207,
      "grad_norm": 0.12929661528379918,
      "learning_rate": 2.5129873473483734e-05,
      "loss": 1.0274,
      "step": 10769
    },
    {
      "epoch": 0.7763281193685576,
      "grad_norm": 0.11445647643921873,
      "learning_rate": 2.5114398511505734e-05,
      "loss": 1.0881,
      "step": 10770
    },
    {
      "epoch": 0.7764002018308945,
      "grad_norm": 0.13087453536026394,
      "learning_rate": 2.5098927631462966e-05,
      "loss": 1.0142,
      "step": 10771
    },
    {
      "epoch": 0.7764722842932315,
      "grad_norm": 0.11989595215378451,
      "learning_rate": 2.5083460834198723e-05,
      "loss": 1.034,
      "step": 10772
    },
    {
      "epoch": 0.7765443667555684,
      "grad_norm": 0.15454965030091228,
      "learning_rate": 2.506799812055609e-05,
      "loss": 0.9987,
      "step": 10773
    },
    {
      "epoch": 0.7766164492179053,
      "grad_norm": 0.10865739904147387,
      "learning_rate": 2.5052539491377912e-05,
      "loss": 0.9706,
      "step": 10774
    },
    {
      "epoch": 0.7766885316802422,
      "grad_norm": 0.09914620242098135,
      "learning_rate": 2.503708494750687e-05,
      "loss": 0.9535,
      "step": 10775
    },
    {
      "epoch": 0.7767606141425791,
      "grad_norm": 0.12198895920216372,
      "learning_rate": 2.5021634489785273e-05,
      "loss": 1.0886,
      "step": 10776
    },
    {
      "epoch": 0.776832696604916,
      "grad_norm": 0.11631065936532231,
      "learning_rate": 2.5006188119055405e-05,
      "loss": 1.1267,
      "step": 10777
    },
    {
      "epoch": 0.776904779067253,
      "grad_norm": 0.26382844316018655,
      "learning_rate": 2.499074583615919e-05,
      "loss": 1.0001,
      "step": 10778
    },
    {
      "epoch": 0.7769768615295899,
      "grad_norm": 0.1158711705088626,
      "learning_rate": 2.4975307641938417e-05,
      "loss": 0.9668,
      "step": 10779
    },
    {
      "epoch": 0.7770489439919268,
      "grad_norm": 0.12787058641302546,
      "learning_rate": 2.49598735372345e-05,
      "loss": 1.1161,
      "step": 10780
    },
    {
      "epoch": 0.7771210264542637,
      "grad_norm": 0.12172376150139967,
      "learning_rate": 2.494444352288884e-05,
      "loss": 1.1277,
      "step": 10781
    },
    {
      "epoch": 0.7771931089166005,
      "grad_norm": 0.14635877837730213,
      "learning_rate": 2.4929017599742476e-05,
      "loss": 1.1368,
      "step": 10782
    },
    {
      "epoch": 0.7772651913789375,
      "grad_norm": 0.12545983660218415,
      "learning_rate": 2.4913595768636278e-05,
      "loss": 1.029,
      "step": 10783
    },
    {
      "epoch": 0.7773372738412744,
      "grad_norm": 0.12473415535073124,
      "learning_rate": 2.4898178030410836e-05,
      "loss": 1.0189,
      "step": 10784
    },
    {
      "epoch": 0.7774093563036113,
      "grad_norm": 0.12239938096888471,
      "learning_rate": 2.488276438590653e-05,
      "loss": 1.0014,
      "step": 10785
    },
    {
      "epoch": 0.7774814387659482,
      "grad_norm": 0.12290594043596097,
      "learning_rate": 2.486735483596362e-05,
      "loss": 1.0814,
      "step": 10786
    },
    {
      "epoch": 0.7775535212282851,
      "grad_norm": 0.1169120787972414,
      "learning_rate": 2.4851949381422046e-05,
      "loss": 1.079,
      "step": 10787
    },
    {
      "epoch": 0.7776256036906221,
      "grad_norm": 0.12921296640564098,
      "learning_rate": 2.483654802312151e-05,
      "loss": 1.0259,
      "step": 10788
    },
    {
      "epoch": 0.777697686152959,
      "grad_norm": 0.11594087971137482,
      "learning_rate": 2.4821150761901503e-05,
      "loss": 1.0355,
      "step": 10789
    },
    {
      "epoch": 0.7777697686152959,
      "grad_norm": 0.18507860871166268,
      "learning_rate": 2.4805757598601363e-05,
      "loss": 1.0325,
      "step": 10790
    },
    {
      "epoch": 0.7778418510776328,
      "grad_norm": 0.12505430041854843,
      "learning_rate": 2.4790368534060194e-05,
      "loss": 1.0484,
      "step": 10791
    },
    {
      "epoch": 0.7779139335399697,
      "grad_norm": 0.13150533036284076,
      "learning_rate": 2.4774983569116737e-05,
      "loss": 1.0243,
      "step": 10792
    },
    {
      "epoch": 0.7779860160023067,
      "grad_norm": 0.13560043131421873,
      "learning_rate": 2.4759602704609664e-05,
      "loss": 1.0054,
      "step": 10793
    },
    {
      "epoch": 0.7780580984646436,
      "grad_norm": 0.12804923367292598,
      "learning_rate": 2.474422594137734e-05,
      "loss": 1.0418,
      "step": 10794
    },
    {
      "epoch": 0.7781301809269805,
      "grad_norm": 0.17240902845892106,
      "learning_rate": 2.4728853280258013e-05,
      "loss": 1.1137,
      "step": 10795
    },
    {
      "epoch": 0.7782022633893174,
      "grad_norm": 0.12228247087183544,
      "learning_rate": 2.4713484722089553e-05,
      "loss": 1.0218,
      "step": 10796
    },
    {
      "epoch": 0.7782743458516543,
      "grad_norm": 0.3021972708007548,
      "learning_rate": 2.46981202677097e-05,
      "loss": 1.1268,
      "step": 10797
    },
    {
      "epoch": 0.7783464283139913,
      "grad_norm": 0.11318440051169297,
      "learning_rate": 2.4682759917955932e-05,
      "loss": 0.9824,
      "step": 10798
    },
    {
      "epoch": 0.7784185107763281,
      "grad_norm": 0.11805674313513219,
      "learning_rate": 2.4667403673665623e-05,
      "loss": 1.0075,
      "step": 10799
    },
    {
      "epoch": 0.778490593238665,
      "grad_norm": 0.12302043117986121,
      "learning_rate": 2.4652051535675725e-05,
      "loss": 1.0201,
      "step": 10800
    },
    {
      "epoch": 0.7785626757010019,
      "grad_norm": 0.11181074835986597,
      "learning_rate": 2.46367035048231e-05,
      "loss": 1.0196,
      "step": 10801
    },
    {
      "epoch": 0.7786347581633388,
      "grad_norm": 0.12630370657196052,
      "learning_rate": 2.4621359581944326e-05,
      "loss": 1.0347,
      "step": 10802
    },
    {
      "epoch": 0.7787068406256757,
      "grad_norm": 0.11138358032093776,
      "learning_rate": 2.4606019767875865e-05,
      "loss": 1.0944,
      "step": 10803
    },
    {
      "epoch": 0.7787789230880127,
      "grad_norm": 0.11394423928639337,
      "learning_rate": 2.4590684063453794e-05,
      "loss": 1.042,
      "step": 10804
    },
    {
      "epoch": 0.7788510055503496,
      "grad_norm": 0.11517409152142959,
      "learning_rate": 2.4575352469514066e-05,
      "loss": 1.0263,
      "step": 10805
    },
    {
      "epoch": 0.7789230880126865,
      "grad_norm": 0.16561349342219575,
      "learning_rate": 2.4560024986892406e-05,
      "loss": 0.9822,
      "step": 10806
    },
    {
      "epoch": 0.7789951704750234,
      "grad_norm": 0.13488506010200146,
      "learning_rate": 2.4544701616424292e-05,
      "loss": 1.013,
      "step": 10807
    },
    {
      "epoch": 0.7790672529373603,
      "grad_norm": 0.1293209015528979,
      "learning_rate": 2.4529382358944974e-05,
      "loss": 1.0507,
      "step": 10808
    },
    {
      "epoch": 0.7791393353996973,
      "grad_norm": 0.14195563991511004,
      "learning_rate": 2.4514067215289503e-05,
      "loss": 1.0283,
      "step": 10809
    },
    {
      "epoch": 0.7792114178620342,
      "grad_norm": 0.10900799694237921,
      "learning_rate": 2.4498756186292682e-05,
      "loss": 1.0544,
      "step": 10810
    },
    {
      "epoch": 0.7792835003243711,
      "grad_norm": 0.11246387798682908,
      "learning_rate": 2.4483449272789106e-05,
      "loss": 1.1247,
      "step": 10811
    },
    {
      "epoch": 0.779355582786708,
      "grad_norm": 0.5829309617575026,
      "learning_rate": 2.446814647561313e-05,
      "loss": 1.0297,
      "step": 10812
    },
    {
      "epoch": 0.7794276652490449,
      "grad_norm": 0.13640677735284115,
      "learning_rate": 2.44528477955989e-05,
      "loss": 1.0287,
      "step": 10813
    },
    {
      "epoch": 0.7794997477113819,
      "grad_norm": 0.11897331157256145,
      "learning_rate": 2.4437553233580322e-05,
      "loss": 1.0651,
      "step": 10814
    },
    {
      "epoch": 0.7795718301737188,
      "grad_norm": 0.1661153925700066,
      "learning_rate": 2.4422262790391103e-05,
      "loss": 1.0264,
      "step": 10815
    },
    {
      "epoch": 0.7796439126360557,
      "grad_norm": 0.11848619033448107,
      "learning_rate": 2.4406976466864696e-05,
      "loss": 1.0623,
      "step": 10816
    },
    {
      "epoch": 0.7797159950983925,
      "grad_norm": 0.1412333709169548,
      "learning_rate": 2.4391694263834354e-05,
      "loss": 1.0029,
      "step": 10817
    },
    {
      "epoch": 0.7797880775607294,
      "grad_norm": 0.11831110772887667,
      "learning_rate": 2.4376416182133076e-05,
      "loss": 1.0395,
      "step": 10818
    },
    {
      "epoch": 0.7798601600230664,
      "grad_norm": 0.11820980473147473,
      "learning_rate": 2.436114222259367e-05,
      "loss": 1.0607,
      "step": 10819
    },
    {
      "epoch": 0.7799322424854033,
      "grad_norm": 0.1075574739776452,
      "learning_rate": 2.4345872386048696e-05,
      "loss": 1.0307,
      "step": 10820
    },
    {
      "epoch": 0.7800043249477402,
      "grad_norm": 0.15449423445122062,
      "learning_rate": 2.433060667333049e-05,
      "loss": 1.1116,
      "step": 10821
    },
    {
      "epoch": 0.7800764074100771,
      "grad_norm": 0.11974635242543602,
      "learning_rate": 2.431534508527119e-05,
      "loss": 1.027,
      "step": 10822
    },
    {
      "epoch": 0.780148489872414,
      "grad_norm": 0.2132193835566011,
      "learning_rate": 2.430008762270267e-05,
      "loss": 1.1225,
      "step": 10823
    },
    {
      "epoch": 0.780220572334751,
      "grad_norm": 0.13092547149498127,
      "learning_rate": 2.4284834286456603e-05,
      "loss": 1.0949,
      "step": 10824
    },
    {
      "epoch": 0.7802926547970879,
      "grad_norm": 0.1264864821701768,
      "learning_rate": 2.4269585077364443e-05,
      "loss": 1.0205,
      "step": 10825
    },
    {
      "epoch": 0.7803647372594248,
      "grad_norm": 0.14626795065324133,
      "learning_rate": 2.425433999625739e-05,
      "loss": 1.0794,
      "step": 10826
    },
    {
      "epoch": 0.7804368197217617,
      "grad_norm": 0.12173275839910166,
      "learning_rate": 2.4239099043966462e-05,
      "loss": 1.0071,
      "step": 10827
    },
    {
      "epoch": 0.7805089021840986,
      "grad_norm": 0.14418534532875943,
      "learning_rate": 2.4223862221322424e-05,
      "loss": 1.0276,
      "step": 10828
    },
    {
      "epoch": 0.7805809846464355,
      "grad_norm": 0.11361535397036791,
      "learning_rate": 2.420862952915577e-05,
      "loss": 1.0718,
      "step": 10829
    },
    {
      "epoch": 0.7806530671087725,
      "grad_norm": 0.40180812619157125,
      "learning_rate": 2.4193400968296865e-05,
      "loss": 1.0015,
      "step": 10830
    },
    {
      "epoch": 0.7807251495711094,
      "grad_norm": 0.11741355160111663,
      "learning_rate": 2.4178176539575803e-05,
      "loss": 1.0572,
      "step": 10831
    },
    {
      "epoch": 0.7807972320334463,
      "grad_norm": 0.15674846993372296,
      "learning_rate": 2.4162956243822464e-05,
      "loss": 1.0029,
      "step": 10832
    },
    {
      "epoch": 0.7808693144957832,
      "grad_norm": 0.13605821004002142,
      "learning_rate": 2.414774008186642e-05,
      "loss": 1.0908,
      "step": 10833
    },
    {
      "epoch": 0.78094139695812,
      "grad_norm": 0.13943750685816664,
      "learning_rate": 2.413252805453715e-05,
      "loss": 1.04,
      "step": 10834
    },
    {
      "epoch": 0.781013479420457,
      "grad_norm": 0.1277979295642028,
      "learning_rate": 2.4117320162663836e-05,
      "loss": 1.064,
      "step": 10835
    },
    {
      "epoch": 0.7810855618827939,
      "grad_norm": 0.13210585560411983,
      "learning_rate": 2.410211640707547e-05,
      "loss": 1.0843,
      "step": 10836
    },
    {
      "epoch": 0.7811576443451308,
      "grad_norm": 0.12112420063077022,
      "learning_rate": 2.4086916788600734e-05,
      "loss": 1.0302,
      "step": 10837
    },
    {
      "epoch": 0.7812297268074677,
      "grad_norm": 0.2665143224440453,
      "learning_rate": 2.407172130806814e-05,
      "loss": 1.0729,
      "step": 10838
    },
    {
      "epoch": 0.7813018092698046,
      "grad_norm": 0.10846071818239855,
      "learning_rate": 2.405652996630604e-05,
      "loss": 1.0067,
      "step": 10839
    },
    {
      "epoch": 0.7813738917321416,
      "grad_norm": 0.1264893041390079,
      "learning_rate": 2.4041342764142494e-05,
      "loss": 1.0545,
      "step": 10840
    },
    {
      "epoch": 0.7814459741944785,
      "grad_norm": 0.1268860188595496,
      "learning_rate": 2.4026159702405282e-05,
      "loss": 1.015,
      "step": 10841
    },
    {
      "epoch": 0.7815180566568154,
      "grad_norm": 0.1304565023885869,
      "learning_rate": 2.4010980781922022e-05,
      "loss": 1.0891,
      "step": 10842
    },
    {
      "epoch": 0.7815901391191523,
      "grad_norm": 0.12851681234543758,
      "learning_rate": 2.3995806003520148e-05,
      "loss": 1.0291,
      "step": 10843
    },
    {
      "epoch": 0.7816622215814892,
      "grad_norm": 0.12698368731939316,
      "learning_rate": 2.3980635368026838e-05,
      "loss": 0.9361,
      "step": 10844
    },
    {
      "epoch": 0.7817343040438262,
      "grad_norm": 0.13044817093584554,
      "learning_rate": 2.396546887626896e-05,
      "loss": 1.0513,
      "step": 10845
    },
    {
      "epoch": 0.7818063865061631,
      "grad_norm": 0.1200710455606085,
      "learning_rate": 2.3950306529073218e-05,
      "loss": 1.0469,
      "step": 10846
    },
    {
      "epoch": 0.7818784689685,
      "grad_norm": 0.11782457447067886,
      "learning_rate": 2.3935148327266167e-05,
      "loss": 1.03,
      "step": 10847
    },
    {
      "epoch": 0.7819505514308369,
      "grad_norm": 0.12346953378448842,
      "learning_rate": 2.391999427167404e-05,
      "loss": 1.0833,
      "step": 10848
    },
    {
      "epoch": 0.7820226338931738,
      "grad_norm": 0.11455781656978299,
      "learning_rate": 2.3904844363122847e-05,
      "loss": 1.0534,
      "step": 10849
    },
    {
      "epoch": 0.7820947163555108,
      "grad_norm": 0.12058737789410492,
      "learning_rate": 2.388969860243839e-05,
      "loss": 1.0008,
      "step": 10850
    },
    {
      "epoch": 0.7821667988178476,
      "grad_norm": 0.14198096541433833,
      "learning_rate": 2.387455699044624e-05,
      "loss": 1.0892,
      "step": 10851
    },
    {
      "epoch": 0.7822388812801845,
      "grad_norm": 0.1547502492549071,
      "learning_rate": 2.3859419527971827e-05,
      "loss": 1.0782,
      "step": 10852
    },
    {
      "epoch": 0.7823109637425214,
      "grad_norm": 0.21316591177149163,
      "learning_rate": 2.3844286215840193e-05,
      "loss": 0.9571,
      "step": 10853
    },
    {
      "epoch": 0.7823830462048583,
      "grad_norm": 0.11189085139442335,
      "learning_rate": 2.382915705487626e-05,
      "loss": 1.022,
      "step": 10854
    },
    {
      "epoch": 0.7824551286671952,
      "grad_norm": 0.11050872332282059,
      "learning_rate": 2.38140320459047e-05,
      "loss": 0.9893,
      "step": 10855
    },
    {
      "epoch": 0.7825272111295322,
      "grad_norm": 0.1258252460168354,
      "learning_rate": 2.3798911189750017e-05,
      "loss": 1.0704,
      "step": 10856
    },
    {
      "epoch": 0.7825992935918691,
      "grad_norm": 0.12197995626855068,
      "learning_rate": 2.3783794487236365e-05,
      "loss": 1.0442,
      "step": 10857
    },
    {
      "epoch": 0.782671376054206,
      "grad_norm": 0.11677687989042741,
      "learning_rate": 2.3768681939187763e-05,
      "loss": 1.0025,
      "step": 10858
    },
    {
      "epoch": 0.7827434585165429,
      "grad_norm": 0.12326019736318176,
      "learning_rate": 2.3753573546427976e-05,
      "loss": 0.9951,
      "step": 10859
    },
    {
      "epoch": 0.7828155409788798,
      "grad_norm": 0.10914682440323191,
      "learning_rate": 2.3738469309780554e-05,
      "loss": 1.0571,
      "step": 10860
    },
    {
      "epoch": 0.7828876234412168,
      "grad_norm": 0.12798098281624934,
      "learning_rate": 2.3723369230068815e-05,
      "loss": 1.0258,
      "step": 10861
    },
    {
      "epoch": 0.7829597059035537,
      "grad_norm": 0.10744248678331189,
      "learning_rate": 2.3708273308115835e-05,
      "loss": 0.9303,
      "step": 10862
    },
    {
      "epoch": 0.7830317883658906,
      "grad_norm": 0.11142401573113445,
      "learning_rate": 2.3693181544744493e-05,
      "loss": 1.0111,
      "step": 10863
    },
    {
      "epoch": 0.7831038708282275,
      "grad_norm": 0.9892100199953354,
      "learning_rate": 2.367809394077741e-05,
      "loss": 1.0003,
      "step": 10864
    },
    {
      "epoch": 0.7831759532905644,
      "grad_norm": 0.11370877003145059,
      "learning_rate": 2.3663010497037007e-05,
      "loss": 1.0573,
      "step": 10865
    },
    {
      "epoch": 0.7832480357529014,
      "grad_norm": 1.2950032345513594,
      "learning_rate": 2.3647931214345464e-05,
      "loss": 1.0187,
      "step": 10866
    },
    {
      "epoch": 0.7833201182152383,
      "grad_norm": 0.12065902689041827,
      "learning_rate": 2.3632856093524724e-05,
      "loss": 1.0869,
      "step": 10867
    },
    {
      "epoch": 0.7833922006775752,
      "grad_norm": 0.12288990986387599,
      "learning_rate": 2.361778513539654e-05,
      "loss": 1.0738,
      "step": 10868
    },
    {
      "epoch": 0.783464283139912,
      "grad_norm": 0.1238415967163359,
      "learning_rate": 2.360271834078239e-05,
      "loss": 0.9754,
      "step": 10869
    },
    {
      "epoch": 0.7835363656022489,
      "grad_norm": 0.1167230990833368,
      "learning_rate": 2.3587655710503564e-05,
      "loss": 1.1142,
      "step": 10870
    },
    {
      "epoch": 0.7836084480645858,
      "grad_norm": 0.12624405499279615,
      "learning_rate": 2.3572597245381112e-05,
      "loss": 1.0512,
      "step": 10871
    },
    {
      "epoch": 0.7836805305269228,
      "grad_norm": 0.11610160141493056,
      "learning_rate": 2.3557542946235844e-05,
      "loss": 0.9764,
      "step": 10872
    },
    {
      "epoch": 0.7837526129892597,
      "grad_norm": 0.1400295178477023,
      "learning_rate": 2.354249281388836e-05,
      "loss": 1.1368,
      "step": 10873
    },
    {
      "epoch": 0.7838246954515966,
      "grad_norm": 0.10941457422611624,
      "learning_rate": 2.3527446849159017e-05,
      "loss": 1.0251,
      "step": 10874
    },
    {
      "epoch": 0.7838967779139335,
      "grad_norm": 0.15267850857757345,
      "learning_rate": 2.351240505286798e-05,
      "loss": 1.0323,
      "step": 10875
    },
    {
      "epoch": 0.7839688603762704,
      "grad_norm": 0.1282106887849486,
      "learning_rate": 2.3497367425835138e-05,
      "loss": 1.0621,
      "step": 10876
    },
    {
      "epoch": 0.7840409428386074,
      "grad_norm": 0.1314415168587238,
      "learning_rate": 2.3482333968880176e-05,
      "loss": 0.994,
      "step": 10877
    },
    {
      "epoch": 0.7841130253009443,
      "grad_norm": 0.11393660307557961,
      "learning_rate": 2.3467304682822565e-05,
      "loss": 1.0382,
      "step": 10878
    },
    {
      "epoch": 0.7841851077632812,
      "grad_norm": 0.13672545724414514,
      "learning_rate": 2.345227956848154e-05,
      "loss": 1.0624,
      "step": 10879
    },
    {
      "epoch": 0.7842571902256181,
      "grad_norm": 0.15569125885896226,
      "learning_rate": 2.3437258626676084e-05,
      "loss": 1.031,
      "step": 10880
    },
    {
      "epoch": 0.784329272687955,
      "grad_norm": 0.1254646804490861,
      "learning_rate": 2.3422241858225013e-05,
      "loss": 0.961,
      "step": 10881
    },
    {
      "epoch": 0.784401355150292,
      "grad_norm": 0.12194464530304884,
      "learning_rate": 2.3407229263946784e-05,
      "loss": 1.0432,
      "step": 10882
    },
    {
      "epoch": 0.7844734376126289,
      "grad_norm": 0.26520244517440394,
      "learning_rate": 2.3392220844659806e-05,
      "loss": 0.9354,
      "step": 10883
    },
    {
      "epoch": 0.7845455200749658,
      "grad_norm": 0.11907357848017346,
      "learning_rate": 2.337721660118215e-05,
      "loss": 1.0883,
      "step": 10884
    },
    {
      "epoch": 0.7846176025373027,
      "grad_norm": 0.14252630337352115,
      "learning_rate": 2.3362216534331705e-05,
      "loss": 1.1331,
      "step": 10885
    },
    {
      "epoch": 0.7846896849996395,
      "grad_norm": 0.13689440342361178,
      "learning_rate": 2.3347220644926028e-05,
      "loss": 0.9852,
      "step": 10886
    },
    {
      "epoch": 0.7847617674619765,
      "grad_norm": 0.18794003688112032,
      "learning_rate": 2.3332228933782596e-05,
      "loss": 1.0018,
      "step": 10887
    },
    {
      "epoch": 0.7848338499243134,
      "grad_norm": 0.11814662905816549,
      "learning_rate": 2.3317241401718592e-05,
      "loss": 0.9686,
      "step": 10888
    },
    {
      "epoch": 0.7849059323866503,
      "grad_norm": 0.12439422065640214,
      "learning_rate": 2.3302258049550974e-05,
      "loss": 0.9511,
      "step": 10889
    },
    {
      "epoch": 0.7849780148489872,
      "grad_norm": 0.12769139173773625,
      "learning_rate": 2.3287278878096396e-05,
      "loss": 1.1149,
      "step": 10890
    },
    {
      "epoch": 0.7850500973113241,
      "grad_norm": 0.12337895845115768,
      "learning_rate": 2.3272303888171432e-05,
      "loss": 0.9878,
      "step": 10891
    },
    {
      "epoch": 0.785122179773661,
      "grad_norm": 0.13117724025260546,
      "learning_rate": 2.3257333080592368e-05,
      "loss": 1.0573,
      "step": 10892
    },
    {
      "epoch": 0.785194262235998,
      "grad_norm": 0.7601960595809093,
      "learning_rate": 2.324236645617518e-05,
      "loss": 1.0353,
      "step": 10893
    },
    {
      "epoch": 0.7852663446983349,
      "grad_norm": 0.1379147707420136,
      "learning_rate": 2.322740401573571e-05,
      "loss": 1.121,
      "step": 10894
    },
    {
      "epoch": 0.7853384271606718,
      "grad_norm": 0.13123748625066137,
      "learning_rate": 2.321244576008953e-05,
      "loss": 0.9946,
      "step": 10895
    },
    {
      "epoch": 0.7854105096230087,
      "grad_norm": 0.12693615520902501,
      "learning_rate": 2.3197491690052065e-05,
      "loss": 1.0208,
      "step": 10896
    },
    {
      "epoch": 0.7854825920853457,
      "grad_norm": 0.1807866671753024,
      "learning_rate": 2.3182541806438374e-05,
      "loss": 1.0964,
      "step": 10897
    },
    {
      "epoch": 0.7855546745476826,
      "grad_norm": 0.12388474707980243,
      "learning_rate": 2.3167596110063384e-05,
      "loss": 1.0947,
      "step": 10898
    },
    {
      "epoch": 0.7856267570100195,
      "grad_norm": 0.18127040142458178,
      "learning_rate": 2.315265460174173e-05,
      "loss": 1.0503,
      "step": 10899
    },
    {
      "epoch": 0.7856988394723564,
      "grad_norm": 0.11508110152034699,
      "learning_rate": 2.3137717282287963e-05,
      "loss": 1.1098,
      "step": 10900
    },
    {
      "epoch": 0.7857709219346933,
      "grad_norm": 0.13676348127070043,
      "learning_rate": 2.3122784152516208e-05,
      "loss": 0.9763,
      "step": 10901
    },
    {
      "epoch": 0.7858430043970303,
      "grad_norm": 0.15292095987239926,
      "learning_rate": 2.310785521324047e-05,
      "loss": 1.0976,
      "step": 10902
    },
    {
      "epoch": 0.7859150868593671,
      "grad_norm": 0.10857456491266931,
      "learning_rate": 2.309293046527452e-05,
      "loss": 1.0635,
      "step": 10903
    },
    {
      "epoch": 0.785987169321704,
      "grad_norm": 0.1305287068858355,
      "learning_rate": 2.3078009909431896e-05,
      "loss": 1.0049,
      "step": 10904
    },
    {
      "epoch": 0.7860592517840409,
      "grad_norm": 0.10969332372387008,
      "learning_rate": 2.3063093546525892e-05,
      "loss": 1.1076,
      "step": 10905
    },
    {
      "epoch": 0.7861313342463778,
      "grad_norm": 0.1395549760070026,
      "learning_rate": 2.3048181377369592e-05,
      "loss": 1.0606,
      "step": 10906
    },
    {
      "epoch": 0.7862034167087147,
      "grad_norm": 0.1208817156667878,
      "learning_rate": 2.3033273402775833e-05,
      "loss": 1.0401,
      "step": 10907
    },
    {
      "epoch": 0.7862754991710517,
      "grad_norm": 0.11900479836681699,
      "learning_rate": 2.3018369623557244e-05,
      "loss": 1.0271,
      "step": 10908
    },
    {
      "epoch": 0.7863475816333886,
      "grad_norm": 0.1041536573638938,
      "learning_rate": 2.30034700405262e-05,
      "loss": 1.0858,
      "step": 10909
    },
    {
      "epoch": 0.7864196640957255,
      "grad_norm": 0.11982739657236181,
      "learning_rate": 2.2988574654494875e-05,
      "loss": 1.0201,
      "step": 10910
    },
    {
      "epoch": 0.7864917465580624,
      "grad_norm": 0.1101318250726732,
      "learning_rate": 2.2973683466275188e-05,
      "loss": 0.992,
      "step": 10911
    },
    {
      "epoch": 0.7865638290203993,
      "grad_norm": 0.166188060137314,
      "learning_rate": 2.2958796476678857e-05,
      "loss": 1.0713,
      "step": 10912
    },
    {
      "epoch": 0.7866359114827363,
      "grad_norm": 0.17341086919575965,
      "learning_rate": 2.294391368651735e-05,
      "loss": 0.9836,
      "step": 10913
    },
    {
      "epoch": 0.7867079939450732,
      "grad_norm": 0.1110497098142466,
      "learning_rate": 2.2929035096601905e-05,
      "loss": 1.0691,
      "step": 10914
    },
    {
      "epoch": 0.7867800764074101,
      "grad_norm": 0.11951690402069441,
      "learning_rate": 2.2914160707743538e-05,
      "loss": 1.0864,
      "step": 10915
    },
    {
      "epoch": 0.786852158869747,
      "grad_norm": 0.12524170176195337,
      "learning_rate": 2.2899290520753046e-05,
      "loss": 1.1126,
      "step": 10916
    },
    {
      "epoch": 0.7869242413320839,
      "grad_norm": 0.13366051576849627,
      "learning_rate": 2.2884424536440985e-05,
      "loss": 1.0684,
      "step": 10917
    },
    {
      "epoch": 0.7869963237944209,
      "grad_norm": 0.11420206822282683,
      "learning_rate": 2.286956275561769e-05,
      "loss": 1.0378,
      "step": 10918
    },
    {
      "epoch": 0.7870684062567578,
      "grad_norm": 0.12314520562407598,
      "learning_rate": 2.2854705179093238e-05,
      "loss": 1.0968,
      "step": 10919
    },
    {
      "epoch": 0.7871404887190947,
      "grad_norm": 0.16579703666008838,
      "learning_rate": 2.2839851807677527e-05,
      "loss": 1.0366,
      "step": 10920
    },
    {
      "epoch": 0.7872125711814315,
      "grad_norm": 0.12577681365858334,
      "learning_rate": 2.2825002642180182e-05,
      "loss": 1.0784,
      "step": 10921
    },
    {
      "epoch": 0.7872846536437684,
      "grad_norm": 0.15742014406500326,
      "learning_rate": 2.2810157683410627e-05,
      "loss": 1.0337,
      "step": 10922
    },
    {
      "epoch": 0.7873567361061053,
      "grad_norm": 0.13552198505083435,
      "learning_rate": 2.2795316932178036e-05,
      "loss": 1.1008,
      "step": 10923
    },
    {
      "epoch": 0.7874288185684423,
      "grad_norm": 0.11536714719222088,
      "learning_rate": 2.278048038929136e-05,
      "loss": 1.0421,
      "step": 10924
    },
    {
      "epoch": 0.7875009010307792,
      "grad_norm": 0.13799723664854438,
      "learning_rate": 2.2765648055559353e-05,
      "loss": 0.9709,
      "step": 10925
    },
    {
      "epoch": 0.7875729834931161,
      "grad_norm": 0.16965184981107978,
      "learning_rate": 2.2750819931790436e-05,
      "loss": 1.0348,
      "step": 10926
    },
    {
      "epoch": 0.787645065955453,
      "grad_norm": 0.13352481576511002,
      "learning_rate": 2.273599601879295e-05,
      "loss": 1.013,
      "step": 10927
    },
    {
      "epoch": 0.78771714841779,
      "grad_norm": 0.11110771520343339,
      "learning_rate": 2.2721176317374903e-05,
      "loss": 0.9644,
      "step": 10928
    },
    {
      "epoch": 0.7877892308801269,
      "grad_norm": 0.117923228579352,
      "learning_rate": 2.270636082834414e-05,
      "loss": 1.0546,
      "step": 10929
    },
    {
      "epoch": 0.7878613133424638,
      "grad_norm": 0.11764816273628542,
      "learning_rate": 2.2691549552508128e-05,
      "loss": 1.0022,
      "step": 10930
    },
    {
      "epoch": 0.7879333958048007,
      "grad_norm": 0.11398257697695247,
      "learning_rate": 2.267674249067433e-05,
      "loss": 1.0534,
      "step": 10931
    },
    {
      "epoch": 0.7880054782671376,
      "grad_norm": 0.15188841431601466,
      "learning_rate": 2.266193964364981e-05,
      "loss": 1.0366,
      "step": 10932
    },
    {
      "epoch": 0.7880775607294745,
      "grad_norm": 0.12570486644396006,
      "learning_rate": 2.26471410122415e-05,
      "loss": 1.0542,
      "step": 10933
    },
    {
      "epoch": 0.7881496431918115,
      "grad_norm": 0.1194472891498197,
      "learning_rate": 2.263234659725596e-05,
      "loss": 1.0864,
      "step": 10934
    },
    {
      "epoch": 0.7882217256541484,
      "grad_norm": 0.16461381616946516,
      "learning_rate": 2.2617556399499716e-05,
      "loss": 1.0491,
      "step": 10935
    },
    {
      "epoch": 0.7882938081164853,
      "grad_norm": 0.12392824583535345,
      "learning_rate": 2.260277041977893e-05,
      "loss": 0.9387,
      "step": 10936
    },
    {
      "epoch": 0.7883658905788222,
      "grad_norm": 0.11860097352549767,
      "learning_rate": 2.2587988658899606e-05,
      "loss": 1.0573,
      "step": 10937
    },
    {
      "epoch": 0.788437973041159,
      "grad_norm": 0.1271898591633138,
      "learning_rate": 2.2573211117667404e-05,
      "loss": 0.9949,
      "step": 10938
    },
    {
      "epoch": 0.788510055503496,
      "grad_norm": 0.12678697338877506,
      "learning_rate": 2.2558437796887853e-05,
      "loss": 1.0649,
      "step": 10939
    },
    {
      "epoch": 0.7885821379658329,
      "grad_norm": 0.13023891719454933,
      "learning_rate": 2.2543668697366293e-05,
      "loss": 0.9491,
      "step": 10940
    },
    {
      "epoch": 0.7886542204281698,
      "grad_norm": 0.12019582547866618,
      "learning_rate": 2.2528903819907752e-05,
      "loss": 0.9752,
      "step": 10941
    },
    {
      "epoch": 0.7887263028905067,
      "grad_norm": 0.13759091218027403,
      "learning_rate": 2.2514143165317002e-05,
      "loss": 1.0188,
      "step": 10942
    },
    {
      "epoch": 0.7887983853528436,
      "grad_norm": 0.1262355580772203,
      "learning_rate": 2.249938673439863e-05,
      "loss": 1.0,
      "step": 10943
    },
    {
      "epoch": 0.7888704678151806,
      "grad_norm": 0.19689904130161928,
      "learning_rate": 2.248463452795705e-05,
      "loss": 1.0807,
      "step": 10944
    },
    {
      "epoch": 0.7889425502775175,
      "grad_norm": 0.12017238293066135,
      "learning_rate": 2.246988654679639e-05,
      "loss": 1.055,
      "step": 10945
    },
    {
      "epoch": 0.7890146327398544,
      "grad_norm": 0.14149341737202706,
      "learning_rate": 2.2455142791720486e-05,
      "loss": 1.0421,
      "step": 10946
    },
    {
      "epoch": 0.7890867152021913,
      "grad_norm": 0.15401982304821243,
      "learning_rate": 2.244040326353304e-05,
      "loss": 0.9677,
      "step": 10947
    },
    {
      "epoch": 0.7891587976645282,
      "grad_norm": 0.12702864786724988,
      "learning_rate": 2.2425667963037454e-05,
      "loss": 1.0445,
      "step": 10948
    },
    {
      "epoch": 0.7892308801268652,
      "grad_norm": 0.12038088890382033,
      "learning_rate": 2.2410936891037015e-05,
      "loss": 1.0438,
      "step": 10949
    },
    {
      "epoch": 0.7893029625892021,
      "grad_norm": 0.11149788547857362,
      "learning_rate": 2.2396210048334622e-05,
      "loss": 1.0199,
      "step": 10950
    },
    {
      "epoch": 0.789375045051539,
      "grad_norm": 0.10836173930198019,
      "learning_rate": 2.238148743573305e-05,
      "loss": 1.0093,
      "step": 10951
    },
    {
      "epoch": 0.7894471275138759,
      "grad_norm": 0.11335017683652834,
      "learning_rate": 2.2366769054034775e-05,
      "loss": 1.0458,
      "step": 10952
    },
    {
      "epoch": 0.7895192099762128,
      "grad_norm": 0.11140778835636732,
      "learning_rate": 2.2352054904042174e-05,
      "loss": 1.0048,
      "step": 10953
    },
    {
      "epoch": 0.7895912924385498,
      "grad_norm": 0.13193088701206762,
      "learning_rate": 2.2337344986557208e-05,
      "loss": 1.0786,
      "step": 10954
    },
    {
      "epoch": 0.7896633749008866,
      "grad_norm": 0.1257834210699857,
      "learning_rate": 2.232263930238172e-05,
      "loss": 1.0491,
      "step": 10955
    },
    {
      "epoch": 0.7897354573632235,
      "grad_norm": 0.1210262385151333,
      "learning_rate": 2.2307937852317296e-05,
      "loss": 0.986,
      "step": 10956
    },
    {
      "epoch": 0.7898075398255604,
      "grad_norm": 0.1276222890030091,
      "learning_rate": 2.229324063716537e-05,
      "loss": 1.0426,
      "step": 10957
    },
    {
      "epoch": 0.7898796222878973,
      "grad_norm": 0.11659106425626899,
      "learning_rate": 2.2278547657726978e-05,
      "loss": 1.0686,
      "step": 10958
    },
    {
      "epoch": 0.7899517047502342,
      "grad_norm": 0.11251418859829986,
      "learning_rate": 2.226385891480306e-05,
      "loss": 1.0721,
      "step": 10959
    },
    {
      "epoch": 0.7900237872125712,
      "grad_norm": 0.13789334407132808,
      "learning_rate": 2.2249174409194274e-05,
      "loss": 1.0354,
      "step": 10960
    },
    {
      "epoch": 0.7900958696749081,
      "grad_norm": 0.11820984344397935,
      "learning_rate": 2.223449414170107e-05,
      "loss": 1.0554,
      "step": 10961
    },
    {
      "epoch": 0.790167952137245,
      "grad_norm": 0.11532481802984058,
      "learning_rate": 2.2219818113123635e-05,
      "loss": 0.9568,
      "step": 10962
    },
    {
      "epoch": 0.7902400345995819,
      "grad_norm": 0.7795227838782821,
      "learning_rate": 2.220514632426196e-05,
      "loss": 0.9849,
      "step": 10963
    },
    {
      "epoch": 0.7903121170619188,
      "grad_norm": 0.14171290800420427,
      "learning_rate": 2.219047877591578e-05,
      "loss": 1.0865,
      "step": 10964
    },
    {
      "epoch": 0.7903841995242558,
      "grad_norm": 0.13136450606865407,
      "learning_rate": 2.217581546888461e-05,
      "loss": 1.1093,
      "step": 10965
    },
    {
      "epoch": 0.7904562819865927,
      "grad_norm": 0.12796640829525463,
      "learning_rate": 2.2161156403967733e-05,
      "loss": 0.9914,
      "step": 10966
    },
    {
      "epoch": 0.7905283644489296,
      "grad_norm": 0.13706747875555714,
      "learning_rate": 2.2146501581964196e-05,
      "loss": 1.0456,
      "step": 10967
    },
    {
      "epoch": 0.7906004469112665,
      "grad_norm": 0.12024700592107894,
      "learning_rate": 2.213185100367282e-05,
      "loss": 1.0117,
      "step": 10968
    },
    {
      "epoch": 0.7906725293736034,
      "grad_norm": 0.11439410886712628,
      "learning_rate": 2.2117204669892198e-05,
      "loss": 1.016,
      "step": 10969
    },
    {
      "epoch": 0.7907446118359404,
      "grad_norm": 0.12038939785054346,
      "learning_rate": 2.210256258142068e-05,
      "loss": 1.065,
      "step": 10970
    },
    {
      "epoch": 0.7908166942982773,
      "grad_norm": 0.28459054669485484,
      "learning_rate": 2.2087924739056387e-05,
      "loss": 1.0675,
      "step": 10971
    },
    {
      "epoch": 0.7908887767606141,
      "grad_norm": 0.12557090980325944,
      "learning_rate": 2.207329114359723e-05,
      "loss": 0.9974,
      "step": 10972
    },
    {
      "epoch": 0.790960859222951,
      "grad_norm": 0.11254852838249571,
      "learning_rate": 2.205866179584084e-05,
      "loss": 0.9533,
      "step": 10973
    },
    {
      "epoch": 0.7910329416852879,
      "grad_norm": 0.13165571733848197,
      "learning_rate": 2.204403669658468e-05,
      "loss": 1.1028,
      "step": 10974
    },
    {
      "epoch": 0.7911050241476248,
      "grad_norm": 0.11654797903126417,
      "learning_rate": 2.202941584662592e-05,
      "loss": 1.0846,
      "step": 10975
    },
    {
      "epoch": 0.7911771066099618,
      "grad_norm": 0.12051739935083741,
      "learning_rate": 2.2014799246761553e-05,
      "loss": 1.0571,
      "step": 10976
    },
    {
      "epoch": 0.7912491890722987,
      "grad_norm": 0.11326902764005607,
      "learning_rate": 2.2000186897788298e-05,
      "loss": 1.079,
      "step": 10977
    },
    {
      "epoch": 0.7913212715346356,
      "grad_norm": 0.11208252485252222,
      "learning_rate": 2.198557880050266e-05,
      "loss": 1.0719,
      "step": 10978
    },
    {
      "epoch": 0.7913933539969725,
      "grad_norm": 0.16899959947362533,
      "learning_rate": 2.1970974955700917e-05,
      "loss": 1.0644,
      "step": 10979
    },
    {
      "epoch": 0.7914654364593094,
      "grad_norm": 0.14315137668957623,
      "learning_rate": 2.1956375364179116e-05,
      "loss": 1.0391,
      "step": 10980
    },
    {
      "epoch": 0.7915375189216464,
      "grad_norm": 0.1284732062633046,
      "learning_rate": 2.194178002673304e-05,
      "loss": 1.0899,
      "step": 10981
    },
    {
      "epoch": 0.7916096013839833,
      "grad_norm": 0.12083729933372413,
      "learning_rate": 2.1927188944158315e-05,
      "loss": 0.9721,
      "step": 10982
    },
    {
      "epoch": 0.7916816838463202,
      "grad_norm": 0.12463113670360783,
      "learning_rate": 2.1912602117250203e-05,
      "loss": 1.0235,
      "step": 10983
    },
    {
      "epoch": 0.7917537663086571,
      "grad_norm": 0.11667433771620209,
      "learning_rate": 2.1898019546803882e-05,
      "loss": 0.999,
      "step": 10984
    },
    {
      "epoch": 0.791825848770994,
      "grad_norm": 0.11266459098479432,
      "learning_rate": 2.188344123361423e-05,
      "loss": 1.0177,
      "step": 10985
    },
    {
      "epoch": 0.791897931233331,
      "grad_norm": 0.1566214452985663,
      "learning_rate": 2.1868867178475893e-05,
      "loss": 1.0597,
      "step": 10986
    },
    {
      "epoch": 0.7919700136956679,
      "grad_norm": 0.13618309014175548,
      "learning_rate": 2.185429738218323e-05,
      "loss": 1.0075,
      "step": 10987
    },
    {
      "epoch": 0.7920420961580048,
      "grad_norm": 0.21372019125521308,
      "learning_rate": 2.18397318455305e-05,
      "loss": 1.1046,
      "step": 10988
    },
    {
      "epoch": 0.7921141786203417,
      "grad_norm": 0.12663310272752148,
      "learning_rate": 2.1825170569311627e-05,
      "loss": 1.1049,
      "step": 10989
    },
    {
      "epoch": 0.7921862610826785,
      "grad_norm": 0.11510453123117162,
      "learning_rate": 2.181061355432036e-05,
      "loss": 0.9788,
      "step": 10990
    },
    {
      "epoch": 0.7922583435450155,
      "grad_norm": 0.12472323069443537,
      "learning_rate": 2.1796060801350116e-05,
      "loss": 1.0618,
      "step": 10991
    },
    {
      "epoch": 0.7923304260073524,
      "grad_norm": 0.1314380022573867,
      "learning_rate": 2.1781512311194173e-05,
      "loss": 1.1844,
      "step": 10992
    },
    {
      "epoch": 0.7924025084696893,
      "grad_norm": 0.19726052225246865,
      "learning_rate": 2.176696808464559e-05,
      "loss": 1.0605,
      "step": 10993
    },
    {
      "epoch": 0.7924745909320262,
      "grad_norm": 0.13449640599363524,
      "learning_rate": 2.1752428122497182e-05,
      "loss": 0.9998,
      "step": 10994
    },
    {
      "epoch": 0.7925466733943631,
      "grad_norm": 0.14268070472111746,
      "learning_rate": 2.1737892425541416e-05,
      "loss": 1.113,
      "step": 10995
    },
    {
      "epoch": 0.7926187558567,
      "grad_norm": 0.1328911341399347,
      "learning_rate": 2.172336099457065e-05,
      "loss": 1.0404,
      "step": 10996
    },
    {
      "epoch": 0.792690838319037,
      "grad_norm": 0.19067541187869308,
      "learning_rate": 2.170883383037703e-05,
      "loss": 1.0819,
      "step": 10997
    },
    {
      "epoch": 0.7927629207813739,
      "grad_norm": 0.12482883508951094,
      "learning_rate": 2.1694310933752395e-05,
      "loss": 1.0608,
      "step": 10998
    },
    {
      "epoch": 0.7928350032437108,
      "grad_norm": 0.1391075625136789,
      "learning_rate": 2.167979230548832e-05,
      "loss": 0.9942,
      "step": 10999
    },
    {
      "epoch": 0.7929070857060477,
      "grad_norm": 0.1702584809743685,
      "learning_rate": 2.166527794637623e-05,
      "loss": 0.9149,
      "step": 11000
    },
    {
      "epoch": 0.7929791681683847,
      "grad_norm": 0.12029284809958286,
      "learning_rate": 2.165076785720731e-05,
      "loss": 0.9942,
      "step": 11001
    },
    {
      "epoch": 0.7930512506307216,
      "grad_norm": 0.11759375124764762,
      "learning_rate": 2.1636262038772504e-05,
      "loss": 1.0152,
      "step": 11002
    },
    {
      "epoch": 0.7931233330930585,
      "grad_norm": 0.13734284555791443,
      "learning_rate": 2.162176049186245e-05,
      "loss": 1.0042,
      "step": 11003
    },
    {
      "epoch": 0.7931954155553954,
      "grad_norm": 0.12012246171321203,
      "learning_rate": 2.160726321726764e-05,
      "loss": 1.1035,
      "step": 11004
    },
    {
      "epoch": 0.7932674980177323,
      "grad_norm": 0.1248221116404457,
      "learning_rate": 2.159277021577828e-05,
      "loss": 0.9985,
      "step": 11005
    },
    {
      "epoch": 0.7933395804800693,
      "grad_norm": 0.12224005393774579,
      "learning_rate": 2.157828148818445e-05,
      "loss": 1.0654,
      "step": 11006
    },
    {
      "epoch": 0.7934116629424061,
      "grad_norm": 0.1549423801463581,
      "learning_rate": 2.1563797035275836e-05,
      "loss": 1.0723,
      "step": 11007
    },
    {
      "epoch": 0.793483745404743,
      "grad_norm": 0.20705682544625834,
      "learning_rate": 2.1549316857841993e-05,
      "loss": 1.0234,
      "step": 11008
    },
    {
      "epoch": 0.7935558278670799,
      "grad_norm": 0.11891599640302573,
      "learning_rate": 2.1534840956672186e-05,
      "loss": 1.0344,
      "step": 11009
    },
    {
      "epoch": 0.7936279103294168,
      "grad_norm": 0.13526297221574107,
      "learning_rate": 2.1520369332555577e-05,
      "loss": 1.0961,
      "step": 11010
    },
    {
      "epoch": 0.7936999927917537,
      "grad_norm": 0.12158427094746159,
      "learning_rate": 2.1505901986280908e-05,
      "loss": 0.9997,
      "step": 11011
    },
    {
      "epoch": 0.7937720752540907,
      "grad_norm": 0.11274612776471482,
      "learning_rate": 2.149143891863681e-05,
      "loss": 1.1044,
      "step": 11012
    },
    {
      "epoch": 0.7938441577164276,
      "grad_norm": 0.13281108959421636,
      "learning_rate": 2.147698013041165e-05,
      "loss": 1.008,
      "step": 11013
    },
    {
      "epoch": 0.7939162401787645,
      "grad_norm": 0.15112515313300562,
      "learning_rate": 2.146252562239356e-05,
      "loss": 1.0435,
      "step": 11014
    },
    {
      "epoch": 0.7939883226411014,
      "grad_norm": 0.11766405783165586,
      "learning_rate": 2.1448075395370437e-05,
      "loss": 1.0008,
      "step": 11015
    },
    {
      "epoch": 0.7940604051034383,
      "grad_norm": 0.1298439035280954,
      "learning_rate": 2.1433629450129954e-05,
      "loss": 1.0986,
      "step": 11016
    },
    {
      "epoch": 0.7941324875657753,
      "grad_norm": 0.12301576265355854,
      "learning_rate": 2.141918778745954e-05,
      "loss": 1.0388,
      "step": 11017
    },
    {
      "epoch": 0.7942045700281122,
      "grad_norm": 0.11215264856539162,
      "learning_rate": 2.140475040814639e-05,
      "loss": 1.0913,
      "step": 11018
    },
    {
      "epoch": 0.7942766524904491,
      "grad_norm": 0.11917988563868719,
      "learning_rate": 2.1390317312977482e-05,
      "loss": 1.1286,
      "step": 11019
    },
    {
      "epoch": 0.794348734952786,
      "grad_norm": 0.14004398138483465,
      "learning_rate": 2.137588850273955e-05,
      "loss": 1.0569,
      "step": 11020
    },
    {
      "epoch": 0.7944208174151229,
      "grad_norm": 0.13249352810361018,
      "learning_rate": 2.1361463978219077e-05,
      "loss": 1.04,
      "step": 11021
    },
    {
      "epoch": 0.7944928998774599,
      "grad_norm": 0.12441119579965829,
      "learning_rate": 2.134704374020233e-05,
      "loss": 1.0303,
      "step": 11022
    },
    {
      "epoch": 0.7945649823397968,
      "grad_norm": 0.3952088926106316,
      "learning_rate": 2.1332627789475358e-05,
      "loss": 1.0607,
      "step": 11023
    },
    {
      "epoch": 0.7946370648021336,
      "grad_norm": 0.11088905795296057,
      "learning_rate": 2.1318216126823952e-05,
      "loss": 1.1532,
      "step": 11024
    },
    {
      "epoch": 0.7947091472644705,
      "grad_norm": 0.12815804039564307,
      "learning_rate": 2.1303808753033672e-05,
      "loss": 0.9936,
      "step": 11025
    },
    {
      "epoch": 0.7947812297268074,
      "grad_norm": 0.11737911224926821,
      "learning_rate": 2.128940566888985e-05,
      "loss": 0.9271,
      "step": 11026
    },
    {
      "epoch": 0.7948533121891443,
      "grad_norm": 1.1980939589552566,
      "learning_rate": 2.1275006875177584e-05,
      "loss": 1.0254,
      "step": 11027
    },
    {
      "epoch": 0.7949253946514813,
      "grad_norm": 0.13878616309550337,
      "learning_rate": 2.1260612372681744e-05,
      "loss": 1.0874,
      "step": 11028
    },
    {
      "epoch": 0.7949974771138182,
      "grad_norm": 0.1330636574659431,
      "learning_rate": 2.1246222162186935e-05,
      "loss": 1.0013,
      "step": 11029
    },
    {
      "epoch": 0.7950695595761551,
      "grad_norm": 0.1247692530866748,
      "learning_rate": 2.123183624447759e-05,
      "loss": 0.9647,
      "step": 11030
    },
    {
      "epoch": 0.795141642038492,
      "grad_norm": 0.11903223766794156,
      "learning_rate": 2.121745462033784e-05,
      "loss": 1.0136,
      "step": 11031
    },
    {
      "epoch": 0.795213724500829,
      "grad_norm": 0.1244229564741809,
      "learning_rate": 2.120307729055162e-05,
      "loss": 1.0797,
      "step": 11032
    },
    {
      "epoch": 0.7952858069631659,
      "grad_norm": 0.10982095317139577,
      "learning_rate": 2.1188704255902626e-05,
      "loss": 0.9933,
      "step": 11033
    },
    {
      "epoch": 0.7953578894255028,
      "grad_norm": 0.12379509299796464,
      "learning_rate": 2.1174335517174317e-05,
      "loss": 1.0291,
      "step": 11034
    },
    {
      "epoch": 0.7954299718878397,
      "grad_norm": 0.11613014659899613,
      "learning_rate": 2.1159971075149944e-05,
      "loss": 1.1142,
      "step": 11035
    },
    {
      "epoch": 0.7955020543501766,
      "grad_norm": 0.11052576429118115,
      "learning_rate": 2.1145610930612413e-05,
      "loss": 1.0148,
      "step": 11036
    },
    {
      "epoch": 0.7955741368125135,
      "grad_norm": 0.11220753725778791,
      "learning_rate": 2.113125508434457e-05,
      "loss": 1.0118,
      "step": 11037
    },
    {
      "epoch": 0.7956462192748505,
      "grad_norm": 0.15333596563887045,
      "learning_rate": 2.1116903537128907e-05,
      "loss": 1.0588,
      "step": 11038
    },
    {
      "epoch": 0.7957183017371874,
      "grad_norm": 0.11384473716238715,
      "learning_rate": 2.110255628974773e-05,
      "loss": 1.0663,
      "step": 11039
    },
    {
      "epoch": 0.7957903841995243,
      "grad_norm": 0.10823753439960514,
      "learning_rate": 2.1088213342983022e-05,
      "loss": 0.9921,
      "step": 11040
    },
    {
      "epoch": 0.7958624666618612,
      "grad_norm": 0.10081328725994775,
      "learning_rate": 2.107387469761668e-05,
      "loss": 0.9317,
      "step": 11041
    },
    {
      "epoch": 0.795934549124198,
      "grad_norm": 0.1288992392731102,
      "learning_rate": 2.1059540354430264e-05,
      "loss": 1.1206,
      "step": 11042
    },
    {
      "epoch": 0.796006631586535,
      "grad_norm": 0.1748287664276268,
      "learning_rate": 2.1045210314205134e-05,
      "loss": 1.0838,
      "step": 11043
    },
    {
      "epoch": 0.7960787140488719,
      "grad_norm": 0.12060479052959043,
      "learning_rate": 2.1030884577722344e-05,
      "loss": 1.0628,
      "step": 11044
    },
    {
      "epoch": 0.7961507965112088,
      "grad_norm": 0.13419204752889374,
      "learning_rate": 2.1016563145762857e-05,
      "loss": 1.0237,
      "step": 11045
    },
    {
      "epoch": 0.7962228789735457,
      "grad_norm": 0.11548634923723528,
      "learning_rate": 2.1002246019107308e-05,
      "loss": 0.9711,
      "step": 11046
    },
    {
      "epoch": 0.7962949614358826,
      "grad_norm": 0.13127237205548753,
      "learning_rate": 2.098793319853605e-05,
      "loss": 0.9874,
      "step": 11047
    },
    {
      "epoch": 0.7963670438982196,
      "grad_norm": 0.12687735042277692,
      "learning_rate": 2.0973624684829296e-05,
      "loss": 0.9808,
      "step": 11048
    },
    {
      "epoch": 0.7964391263605565,
      "grad_norm": 0.14922146320891827,
      "learning_rate": 2.0959320478766965e-05,
      "loss": 0.9746,
      "step": 11049
    },
    {
      "epoch": 0.7965112088228934,
      "grad_norm": 0.11800878360625641,
      "learning_rate": 2.094502058112884e-05,
      "loss": 1.0024,
      "step": 11050
    },
    {
      "epoch": 0.7965832912852303,
      "grad_norm": 0.21013505430694804,
      "learning_rate": 2.0930724992694318e-05,
      "loss": 1.0159,
      "step": 11051
    },
    {
      "epoch": 0.7966553737475672,
      "grad_norm": 0.12088963768974414,
      "learning_rate": 2.0916433714242646e-05,
      "loss": 1.0864,
      "step": 11052
    },
    {
      "epoch": 0.7967274562099042,
      "grad_norm": 0.11822952376244847,
      "learning_rate": 2.090214674655282e-05,
      "loss": 0.9301,
      "step": 11053
    },
    {
      "epoch": 0.7967995386722411,
      "grad_norm": 0.11804750531501443,
      "learning_rate": 2.0887864090403676e-05,
      "loss": 1.078,
      "step": 11054
    },
    {
      "epoch": 0.796871621134578,
      "grad_norm": 0.11377261842724772,
      "learning_rate": 2.0873585746573666e-05,
      "loss": 1.0233,
      "step": 11055
    },
    {
      "epoch": 0.7969437035969149,
      "grad_norm": 0.14315448707020698,
      "learning_rate": 2.0859311715841122e-05,
      "loss": 1.0004,
      "step": 11056
    },
    {
      "epoch": 0.7970157860592518,
      "grad_norm": 0.12399469917512726,
      "learning_rate": 2.0845041998984105e-05,
      "loss": 0.9587,
      "step": 11057
    },
    {
      "epoch": 0.7970878685215887,
      "grad_norm": 0.15493726215595677,
      "learning_rate": 2.0830776596780432e-05,
      "loss": 1.0212,
      "step": 11058
    },
    {
      "epoch": 0.7971599509839256,
      "grad_norm": 0.14677547176934025,
      "learning_rate": 2.0816515510007707e-05,
      "loss": 1.1007,
      "step": 11059
    },
    {
      "epoch": 0.7972320334462625,
      "grad_norm": 0.13564195332219559,
      "learning_rate": 2.080225873944328e-05,
      "loss": 1.1025,
      "step": 11060
    },
    {
      "epoch": 0.7973041159085994,
      "grad_norm": 0.12434768452197129,
      "learning_rate": 2.078800628586428e-05,
      "loss": 1.0596,
      "step": 11061
    },
    {
      "epoch": 0.7973761983709363,
      "grad_norm": 0.1151950535622127,
      "learning_rate": 2.0773758150047583e-05,
      "loss": 1.0437,
      "step": 11062
    },
    {
      "epoch": 0.7974482808332732,
      "grad_norm": 0.13346447715113588,
      "learning_rate": 2.0759514332769848e-05,
      "loss": 1.0974,
      "step": 11063
    },
    {
      "epoch": 0.7975203632956102,
      "grad_norm": 0.11069673726557164,
      "learning_rate": 2.0745274834807492e-05,
      "loss": 0.982,
      "step": 11064
    },
    {
      "epoch": 0.7975924457579471,
      "grad_norm": 0.12479033489568654,
      "learning_rate": 2.073103965693668e-05,
      "loss": 1.0132,
      "step": 11065
    },
    {
      "epoch": 0.797664528220284,
      "grad_norm": 0.14132584493732425,
      "learning_rate": 2.0716808799933375e-05,
      "loss": 1.0161,
      "step": 11066
    },
    {
      "epoch": 0.7977366106826209,
      "grad_norm": 1.3048891047533404,
      "learning_rate": 2.0702582264573277e-05,
      "loss": 1.1084,
      "step": 11067
    },
    {
      "epoch": 0.7978086931449578,
      "grad_norm": 0.12055299816151505,
      "learning_rate": 2.068836005163186e-05,
      "loss": 1.0972,
      "step": 11068
    },
    {
      "epoch": 0.7978807756072948,
      "grad_norm": 0.11262065273633791,
      "learning_rate": 2.0674142161884358e-05,
      "loss": 1.0642,
      "step": 11069
    },
    {
      "epoch": 0.7979528580696317,
      "grad_norm": 0.1289140548379718,
      "learning_rate": 2.065992859610578e-05,
      "loss": 1.0165,
      "step": 11070
    },
    {
      "epoch": 0.7980249405319686,
      "grad_norm": 0.12118681307716449,
      "learning_rate": 2.0645719355070902e-05,
      "loss": 1.0577,
      "step": 11071
    },
    {
      "epoch": 0.7980970229943055,
      "grad_norm": 0.1201318132969681,
      "learning_rate": 2.063151443955422e-05,
      "loss": 0.9777,
      "step": 11072
    },
    {
      "epoch": 0.7981691054566424,
      "grad_norm": 0.13314931727194526,
      "learning_rate": 2.0617313850330066e-05,
      "loss": 1.1038,
      "step": 11073
    },
    {
      "epoch": 0.7982411879189794,
      "grad_norm": 0.11714857184913798,
      "learning_rate": 2.0603117588172495e-05,
      "loss": 1.033,
      "step": 11074
    },
    {
      "epoch": 0.7983132703813163,
      "grad_norm": 0.13822822501036192,
      "learning_rate": 2.0588925653855307e-05,
      "loss": 1.015,
      "step": 11075
    },
    {
      "epoch": 0.7983853528436531,
      "grad_norm": 0.2873056431864422,
      "learning_rate": 2.0574738048152097e-05,
      "loss": 1.0018,
      "step": 11076
    },
    {
      "epoch": 0.79845743530599,
      "grad_norm": 0.13480509223460554,
      "learning_rate": 2.0560554771836227e-05,
      "loss": 1.0857,
      "step": 11077
    },
    {
      "epoch": 0.7985295177683269,
      "grad_norm": 0.10592700048497226,
      "learning_rate": 2.054637582568082e-05,
      "loss": 0.9978,
      "step": 11078
    },
    {
      "epoch": 0.7986016002306638,
      "grad_norm": 0.1388556789412584,
      "learning_rate": 2.0532201210458755e-05,
      "loss": 1.0698,
      "step": 11079
    },
    {
      "epoch": 0.7986736826930008,
      "grad_norm": 0.11590831710832399,
      "learning_rate": 2.051803092694261e-05,
      "loss": 1.065,
      "step": 11080
    },
    {
      "epoch": 0.7987457651553377,
      "grad_norm": 0.10698753058948984,
      "learning_rate": 2.0503864975904873e-05,
      "loss": 1.0562,
      "step": 11081
    },
    {
      "epoch": 0.7988178476176746,
      "grad_norm": 0.12684431810505012,
      "learning_rate": 2.0489703358117696e-05,
      "loss": 0.9687,
      "step": 11082
    },
    {
      "epoch": 0.7988899300800115,
      "grad_norm": 0.13215337597635118,
      "learning_rate": 2.0475546074353026e-05,
      "loss": 1.0223,
      "step": 11083
    },
    {
      "epoch": 0.7989620125423484,
      "grad_norm": 0.1220946807573336,
      "learning_rate": 2.046139312538249e-05,
      "loss": 0.9744,
      "step": 11084
    },
    {
      "epoch": 0.7990340950046854,
      "grad_norm": 0.11617384470214664,
      "learning_rate": 2.0447244511977627e-05,
      "loss": 1.0657,
      "step": 11085
    },
    {
      "epoch": 0.7991061774670223,
      "grad_norm": 0.11467470362517783,
      "learning_rate": 2.0433100234909642e-05,
      "loss": 1.0393,
      "step": 11086
    },
    {
      "epoch": 0.7991782599293592,
      "grad_norm": 0.11522534245597564,
      "learning_rate": 2.041896029494954e-05,
      "loss": 0.9844,
      "step": 11087
    },
    {
      "epoch": 0.7992503423916961,
      "grad_norm": 0.1388156706524494,
      "learning_rate": 2.040482469286803e-05,
      "loss": 1.0543,
      "step": 11088
    },
    {
      "epoch": 0.799322424854033,
      "grad_norm": 0.11958320361901636,
      "learning_rate": 2.0390693429435627e-05,
      "loss": 1.0663,
      "step": 11089
    },
    {
      "epoch": 0.79939450731637,
      "grad_norm": 0.14019830649801554,
      "learning_rate": 2.0376566505422657e-05,
      "loss": 0.9902,
      "step": 11090
    },
    {
      "epoch": 0.7994665897787069,
      "grad_norm": 0.10151926454286701,
      "learning_rate": 2.0362443921599173e-05,
      "loss": 0.9958,
      "step": 11091
    },
    {
      "epoch": 0.7995386722410438,
      "grad_norm": 0.13092705150919626,
      "learning_rate": 2.0348325678734936e-05,
      "loss": 1.049,
      "step": 11092
    },
    {
      "epoch": 0.7996107547033807,
      "grad_norm": 0.17661872795623101,
      "learning_rate": 2.0334211777599498e-05,
      "loss": 1.0483,
      "step": 11093
    },
    {
      "epoch": 0.7996828371657175,
      "grad_norm": 0.21772972972330135,
      "learning_rate": 2.0320102218962256e-05,
      "loss": 1.0597,
      "step": 11094
    },
    {
      "epoch": 0.7997549196280545,
      "grad_norm": 0.12161708045787717,
      "learning_rate": 2.0305997003592314e-05,
      "loss": 1.0692,
      "step": 11095
    },
    {
      "epoch": 0.7998270020903914,
      "grad_norm": 0.1896776427558219,
      "learning_rate": 2.029189613225847e-05,
      "loss": 1.0741,
      "step": 11096
    },
    {
      "epoch": 0.7998990845527283,
      "grad_norm": 0.11793006303241156,
      "learning_rate": 2.0277799605729342e-05,
      "loss": 1.0999,
      "step": 11097
    },
    {
      "epoch": 0.7999711670150652,
      "grad_norm": 0.14915054974212055,
      "learning_rate": 2.0263707424773382e-05,
      "loss": 1.0512,
      "step": 11098
    },
    {
      "epoch": 0.8000432494774021,
      "grad_norm": 0.13025792909093606,
      "learning_rate": 2.024961959015873e-05,
      "loss": 0.9761,
      "step": 11099
    },
    {
      "epoch": 0.800115331939739,
      "grad_norm": 0.11989538748568919,
      "learning_rate": 2.0235536102653253e-05,
      "loss": 1.0262,
      "step": 11100
    },
    {
      "epoch": 0.800187414402076,
      "grad_norm": 0.13045619909887793,
      "learning_rate": 2.0221456963024656e-05,
      "loss": 0.9825,
      "step": 11101
    },
    {
      "epoch": 0.8002594968644129,
      "grad_norm": 0.14739349092931997,
      "learning_rate": 2.0207382172040347e-05,
      "loss": 0.973,
      "step": 11102
    },
    {
      "epoch": 0.8003315793267498,
      "grad_norm": 0.11927864409737257,
      "learning_rate": 2.0193311730467602e-05,
      "loss": 1.0423,
      "step": 11103
    },
    {
      "epoch": 0.8004036617890867,
      "grad_norm": 0.13056678483221135,
      "learning_rate": 2.017924563907332e-05,
      "loss": 1.0227,
      "step": 11104
    },
    {
      "epoch": 0.8004757442514236,
      "grad_norm": 0.13976339962175224,
      "learning_rate": 2.016518389862424e-05,
      "loss": 0.9945,
      "step": 11105
    },
    {
      "epoch": 0.8005478267137606,
      "grad_norm": 0.11804360374281538,
      "learning_rate": 2.0151126509886843e-05,
      "loss": 0.9528,
      "step": 11106
    },
    {
      "epoch": 0.8006199091760975,
      "grad_norm": 0.12127563144335077,
      "learning_rate": 2.0137073473627444e-05,
      "loss": 0.9827,
      "step": 11107
    },
    {
      "epoch": 0.8006919916384344,
      "grad_norm": 0.11958846742700549,
      "learning_rate": 2.0123024790611987e-05,
      "loss": 0.9649,
      "step": 11108
    },
    {
      "epoch": 0.8007640741007713,
      "grad_norm": 0.15647514607786564,
      "learning_rate": 2.0108980461606287e-05,
      "loss": 1.1285,
      "step": 11109
    },
    {
      "epoch": 0.8008361565631082,
      "grad_norm": 0.10896051327879333,
      "learning_rate": 2.009494048737587e-05,
      "loss": 1.0425,
      "step": 11110
    },
    {
      "epoch": 0.8009082390254451,
      "grad_norm": 0.1264513530772986,
      "learning_rate": 2.0080904868686057e-05,
      "loss": 1.0074,
      "step": 11111
    },
    {
      "epoch": 0.800980321487782,
      "grad_norm": 0.12013424567587336,
      "learning_rate": 2.0066873606301896e-05,
      "loss": 1.0468,
      "step": 11112
    },
    {
      "epoch": 0.8010524039501189,
      "grad_norm": 0.11360737734045526,
      "learning_rate": 2.0052846700988226e-05,
      "loss": 1.0612,
      "step": 11113
    },
    {
      "epoch": 0.8011244864124558,
      "grad_norm": 0.13773133274505103,
      "learning_rate": 2.003882415350965e-05,
      "loss": 1.0977,
      "step": 11114
    },
    {
      "epoch": 0.8011965688747927,
      "grad_norm": 0.11189127371887718,
      "learning_rate": 2.00248059646305e-05,
      "loss": 1.1006,
      "step": 11115
    },
    {
      "epoch": 0.8012686513371297,
      "grad_norm": 0.11422890075005043,
      "learning_rate": 2.0010792135114897e-05,
      "loss": 1.0292,
      "step": 11116
    },
    {
      "epoch": 0.8013407337994666,
      "grad_norm": 0.11322939228419934,
      "learning_rate": 1.9996782665726743e-05,
      "loss": 1.0241,
      "step": 11117
    },
    {
      "epoch": 0.8014128162618035,
      "grad_norm": 0.13912885534165922,
      "learning_rate": 1.998277755722965e-05,
      "loss": 0.9922,
      "step": 11118
    },
    {
      "epoch": 0.8014848987241404,
      "grad_norm": 0.25870539394312003,
      "learning_rate": 1.9968776810387048e-05,
      "loss": 1.0663,
      "step": 11119
    },
    {
      "epoch": 0.8015569811864773,
      "grad_norm": 0.1230209952728383,
      "learning_rate": 1.995478042596207e-05,
      "loss": 1.0551,
      "step": 11120
    },
    {
      "epoch": 0.8016290636488143,
      "grad_norm": 0.10868483500287271,
      "learning_rate": 1.994078840471768e-05,
      "loss": 1.0532,
      "step": 11121
    },
    {
      "epoch": 0.8017011461111512,
      "grad_norm": 0.3163469490956745,
      "learning_rate": 1.992680074741655e-05,
      "loss": 1.0286,
      "step": 11122
    },
    {
      "epoch": 0.8017732285734881,
      "grad_norm": 0.1355128283181825,
      "learning_rate": 1.991281745482113e-05,
      "loss": 1.0814,
      "step": 11123
    },
    {
      "epoch": 0.801845311035825,
      "grad_norm": 0.12471931089222657,
      "learning_rate": 1.9898838527693643e-05,
      "loss": 1.0237,
      "step": 11124
    },
    {
      "epoch": 0.8019173934981619,
      "grad_norm": 0.1483787938009968,
      "learning_rate": 1.9884863966796052e-05,
      "loss": 0.9848,
      "step": 11125
    },
    {
      "epoch": 0.8019894759604989,
      "grad_norm": 0.1347134078668847,
      "learning_rate": 1.987089377289012e-05,
      "loss": 1.0575,
      "step": 11126
    },
    {
      "epoch": 0.8020615584228358,
      "grad_norm": 0.12666760666576807,
      "learning_rate": 1.9856927946737324e-05,
      "loss": 1.0466,
      "step": 11127
    },
    {
      "epoch": 0.8021336408851726,
      "grad_norm": 0.11364824624386884,
      "learning_rate": 1.9842966489098935e-05,
      "loss": 1.0297,
      "step": 11128
    },
    {
      "epoch": 0.8022057233475095,
      "grad_norm": 0.1808919135496906,
      "learning_rate": 1.9829009400735977e-05,
      "loss": 1.0477,
      "step": 11129
    },
    {
      "epoch": 0.8022778058098464,
      "grad_norm": 0.21524328219039934,
      "learning_rate": 1.981505668240924e-05,
      "loss": 0.9657,
      "step": 11130
    },
    {
      "epoch": 0.8023498882721833,
      "grad_norm": 0.13542119093699084,
      "learning_rate": 1.980110833487927e-05,
      "loss": 0.9997,
      "step": 11131
    },
    {
      "epoch": 0.8024219707345203,
      "grad_norm": 0.11267067781396178,
      "learning_rate": 1.9787164358906407e-05,
      "loss": 1.0002,
      "step": 11132
    },
    {
      "epoch": 0.8024940531968572,
      "grad_norm": 0.16432765562766244,
      "learning_rate": 1.977322475525063e-05,
      "loss": 1.0308,
      "step": 11133
    },
    {
      "epoch": 0.8025661356591941,
      "grad_norm": 0.13205835092671192,
      "learning_rate": 1.9759289524671854e-05,
      "loss": 1.0153,
      "step": 11134
    },
    {
      "epoch": 0.802638218121531,
      "grad_norm": 0.11372100926063505,
      "learning_rate": 1.9745358667929658e-05,
      "loss": 0.9747,
      "step": 11135
    },
    {
      "epoch": 0.8027103005838679,
      "grad_norm": 0.12932281398039291,
      "learning_rate": 1.9731432185783426e-05,
      "loss": 0.9583,
      "step": 11136
    },
    {
      "epoch": 0.8027823830462049,
      "grad_norm": 0.11324856065015547,
      "learning_rate": 1.9717510078992185e-05,
      "loss": 0.9878,
      "step": 11137
    },
    {
      "epoch": 0.8028544655085418,
      "grad_norm": 0.11932788145269353,
      "learning_rate": 1.9703592348314903e-05,
      "loss": 1.0521,
      "step": 11138
    },
    {
      "epoch": 0.8029265479708787,
      "grad_norm": 0.11925229279825646,
      "learning_rate": 1.9689678994510184e-05,
      "loss": 1.1225,
      "step": 11139
    },
    {
      "epoch": 0.8029986304332156,
      "grad_norm": 0.12301903094436491,
      "learning_rate": 1.967577001833647e-05,
      "loss": 1.0618,
      "step": 11140
    },
    {
      "epoch": 0.8030707128955525,
      "grad_norm": 0.11521894687130473,
      "learning_rate": 1.9661865420551828e-05,
      "loss": 1.0547,
      "step": 11141
    },
    {
      "epoch": 0.8031427953578895,
      "grad_norm": 0.11960041562247002,
      "learning_rate": 1.9647965201914277e-05,
      "loss": 1.0532,
      "step": 11142
    },
    {
      "epoch": 0.8032148778202264,
      "grad_norm": 0.11943296061647826,
      "learning_rate": 1.9634069363181474e-05,
      "loss": 1.0551,
      "step": 11143
    },
    {
      "epoch": 0.8032869602825633,
      "grad_norm": 0.22942732937757235,
      "learning_rate": 1.9620177905110892e-05,
      "loss": 1.011,
      "step": 11144
    },
    {
      "epoch": 0.8033590427449002,
      "grad_norm": 0.1569483325838822,
      "learning_rate": 1.9606290828459685e-05,
      "loss": 0.9641,
      "step": 11145
    },
    {
      "epoch": 0.803431125207237,
      "grad_norm": 0.11704493909024759,
      "learning_rate": 1.959240813398483e-05,
      "loss": 1.0833,
      "step": 11146
    },
    {
      "epoch": 0.803503207669574,
      "grad_norm": 0.1274404060849779,
      "learning_rate": 1.957852982244309e-05,
      "loss": 1.0195,
      "step": 11147
    },
    {
      "epoch": 0.8035752901319109,
      "grad_norm": 0.12357507742819615,
      "learning_rate": 1.9564655894590987e-05,
      "loss": 1.0156,
      "step": 11148
    },
    {
      "epoch": 0.8036473725942478,
      "grad_norm": 1.1403673794745877,
      "learning_rate": 1.9550786351184703e-05,
      "loss": 0.9791,
      "step": 11149
    },
    {
      "epoch": 0.8037194550565847,
      "grad_norm": 0.12894229263763016,
      "learning_rate": 1.953692119298025e-05,
      "loss": 0.9957,
      "step": 11150
    },
    {
      "epoch": 0.8037915375189216,
      "grad_norm": 0.11886755050737036,
      "learning_rate": 1.9523060420733462e-05,
      "loss": 1.0059,
      "step": 11151
    },
    {
      "epoch": 0.8038636199812585,
      "grad_norm": 0.12745806514096877,
      "learning_rate": 1.950920403519988e-05,
      "loss": 1.0195,
      "step": 11152
    },
    {
      "epoch": 0.8039357024435955,
      "grad_norm": 0.1641984450546065,
      "learning_rate": 1.949535203713474e-05,
      "loss": 1.0496,
      "step": 11153
    },
    {
      "epoch": 0.8040077849059324,
      "grad_norm": 0.10882043745896841,
      "learning_rate": 1.948150442729313e-05,
      "loss": 1.0488,
      "step": 11154
    },
    {
      "epoch": 0.8040798673682693,
      "grad_norm": 0.11889926477013217,
      "learning_rate": 1.9467661206429844e-05,
      "loss": 0.9911,
      "step": 11155
    },
    {
      "epoch": 0.8041519498306062,
      "grad_norm": 0.12936379874882828,
      "learning_rate": 1.9453822375299537e-05,
      "loss": 1.0191,
      "step": 11156
    },
    {
      "epoch": 0.8042240322929431,
      "grad_norm": 0.11239281973787651,
      "learning_rate": 1.9439987934656467e-05,
      "loss": 0.9734,
      "step": 11157
    },
    {
      "epoch": 0.8042961147552801,
      "grad_norm": 0.11920884809611086,
      "learning_rate": 1.942615788525476e-05,
      "loss": 1.0716,
      "step": 11158
    },
    {
      "epoch": 0.804368197217617,
      "grad_norm": 0.1633903790951547,
      "learning_rate": 1.9412332227848263e-05,
      "loss": 1.0173,
      "step": 11159
    },
    {
      "epoch": 0.8044402796799539,
      "grad_norm": 0.12463065562869922,
      "learning_rate": 1.9398510963190676e-05,
      "loss": 0.9431,
      "step": 11160
    },
    {
      "epoch": 0.8045123621422908,
      "grad_norm": 0.26315774628352445,
      "learning_rate": 1.9384694092035294e-05,
      "loss": 1.0336,
      "step": 11161
    },
    {
      "epoch": 0.8045844446046277,
      "grad_norm": 0.12807624994988956,
      "learning_rate": 1.9370881615135285e-05,
      "loss": 0.9923,
      "step": 11162
    },
    {
      "epoch": 0.8046565270669646,
      "grad_norm": 0.7938402012102188,
      "learning_rate": 1.935707353324353e-05,
      "loss": 1.0048,
      "step": 11163
    },
    {
      "epoch": 0.8047286095293015,
      "grad_norm": 0.2520684496111796,
      "learning_rate": 1.9343269847112765e-05,
      "loss": 1.0432,
      "step": 11164
    },
    {
      "epoch": 0.8048006919916384,
      "grad_norm": 0.3211237326986766,
      "learning_rate": 1.9329470557495354e-05,
      "loss": 1.0975,
      "step": 11165
    },
    {
      "epoch": 0.8048727744539753,
      "grad_norm": 0.11689786910995532,
      "learning_rate": 1.9315675665143495e-05,
      "loss": 1.0002,
      "step": 11166
    },
    {
      "epoch": 0.8049448569163122,
      "grad_norm": 0.11825229727142286,
      "learning_rate": 1.930188517080913e-05,
      "loss": 1.0323,
      "step": 11167
    },
    {
      "epoch": 0.8050169393786492,
      "grad_norm": 0.1476711210416349,
      "learning_rate": 1.9288099075243983e-05,
      "loss": 1.0118,
      "step": 11168
    },
    {
      "epoch": 0.8050890218409861,
      "grad_norm": 0.1235536347421427,
      "learning_rate": 1.9274317379199492e-05,
      "loss": 0.9961,
      "step": 11169
    },
    {
      "epoch": 0.805161104303323,
      "grad_norm": 0.1195273908553281,
      "learning_rate": 1.9260540083426903e-05,
      "loss": 1.0464,
      "step": 11170
    },
    {
      "epoch": 0.8052331867656599,
      "grad_norm": 0.11443832604038874,
      "learning_rate": 1.9246767188677195e-05,
      "loss": 1.0456,
      "step": 11171
    },
    {
      "epoch": 0.8053052692279968,
      "grad_norm": 0.1281400160255519,
      "learning_rate": 1.923299869570111e-05,
      "loss": 1.1013,
      "step": 11172
    },
    {
      "epoch": 0.8053773516903338,
      "grad_norm": 0.11450151418447045,
      "learning_rate": 1.9219234605249158e-05,
      "loss": 1.0629,
      "step": 11173
    },
    {
      "epoch": 0.8054494341526707,
      "grad_norm": 0.6946289938945288,
      "learning_rate": 1.9205474918071608e-05,
      "loss": 1.0356,
      "step": 11174
    },
    {
      "epoch": 0.8055215166150076,
      "grad_norm": 0.11264294907927776,
      "learning_rate": 1.9191719634918482e-05,
      "loss": 1.1009,
      "step": 11175
    },
    {
      "epoch": 0.8055935990773445,
      "grad_norm": 0.1354707695993259,
      "learning_rate": 1.9177968756539567e-05,
      "loss": 1.064,
      "step": 11176
    },
    {
      "epoch": 0.8056656815396814,
      "grad_norm": 0.12116489747498294,
      "learning_rate": 1.9164222283684408e-05,
      "loss": 1.0835,
      "step": 11177
    },
    {
      "epoch": 0.8057377640020184,
      "grad_norm": 0.10181048098747371,
      "learning_rate": 1.9150480217102318e-05,
      "loss": 1.0033,
      "step": 11178
    },
    {
      "epoch": 0.8058098464643553,
      "grad_norm": 0.1350575065632151,
      "learning_rate": 1.9136742557542352e-05,
      "loss": 1.0104,
      "step": 11179
    },
    {
      "epoch": 0.8058819289266921,
      "grad_norm": 0.13580680565520356,
      "learning_rate": 1.912300930575335e-05,
      "loss": 1.0535,
      "step": 11180
    },
    {
      "epoch": 0.805954011389029,
      "grad_norm": 0.12416674279014046,
      "learning_rate": 1.910928046248387e-05,
      "loss": 1.0508,
      "step": 11181
    },
    {
      "epoch": 0.8060260938513659,
      "grad_norm": 0.12069478588322907,
      "learning_rate": 1.9095556028482286e-05,
      "loss": 1.0566,
      "step": 11182
    },
    {
      "epoch": 0.8060981763137028,
      "grad_norm": 0.18221045287306842,
      "learning_rate": 1.908183600449669e-05,
      "loss": 1.0265,
      "step": 11183
    },
    {
      "epoch": 0.8061702587760398,
      "grad_norm": 0.12205813284430977,
      "learning_rate": 1.9068120391274936e-05,
      "loss": 1.0346,
      "step": 11184
    },
    {
      "epoch": 0.8062423412383767,
      "grad_norm": 0.1229552920374694,
      "learning_rate": 1.905440918956467e-05,
      "loss": 1.1671,
      "step": 11185
    },
    {
      "epoch": 0.8063144237007136,
      "grad_norm": 0.11614081625106373,
      "learning_rate": 1.9040702400113252e-05,
      "loss": 1.0402,
      "step": 11186
    },
    {
      "epoch": 0.8063865061630505,
      "grad_norm": 0.1352953011877743,
      "learning_rate": 1.902700002366784e-05,
      "loss": 1.0351,
      "step": 11187
    },
    {
      "epoch": 0.8064585886253874,
      "grad_norm": 0.12520019937398344,
      "learning_rate": 1.9013302060975336e-05,
      "loss": 1.031,
      "step": 11188
    },
    {
      "epoch": 0.8065306710877244,
      "grad_norm": 0.11800667570315299,
      "learning_rate": 1.899960851278243e-05,
      "loss": 1.0261,
      "step": 11189
    },
    {
      "epoch": 0.8066027535500613,
      "grad_norm": 0.17253142169202726,
      "learning_rate": 1.8985919379835448e-05,
      "loss": 1.1117,
      "step": 11190
    },
    {
      "epoch": 0.8066748360123982,
      "grad_norm": 0.12249417308160175,
      "learning_rate": 1.897223466288065e-05,
      "loss": 1.0351,
      "step": 11191
    },
    {
      "epoch": 0.8067469184747351,
      "grad_norm": 0.11468827874216045,
      "learning_rate": 1.8958554362663972e-05,
      "loss": 1.0432,
      "step": 11192
    },
    {
      "epoch": 0.806819000937072,
      "grad_norm": 0.12224090025021868,
      "learning_rate": 1.8944878479931126e-05,
      "loss": 1.018,
      "step": 11193
    },
    {
      "epoch": 0.806891083399409,
      "grad_norm": 0.12370065791689158,
      "learning_rate": 1.8931207015427477e-05,
      "loss": 1.0344,
      "step": 11194
    },
    {
      "epoch": 0.8069631658617459,
      "grad_norm": 0.12252176219994018,
      "learning_rate": 1.8917539969898346e-05,
      "loss": 1.1041,
      "step": 11195
    },
    {
      "epoch": 0.8070352483240828,
      "grad_norm": 0.12025678582869674,
      "learning_rate": 1.890387734408867e-05,
      "loss": 1.0417,
      "step": 11196
    },
    {
      "epoch": 0.8071073307864197,
      "grad_norm": 0.15805052617761403,
      "learning_rate": 1.8890219138743203e-05,
      "loss": 0.9336,
      "step": 11197
    },
    {
      "epoch": 0.8071794132487565,
      "grad_norm": 0.13059484303260288,
      "learning_rate": 1.8876565354606402e-05,
      "loss": 1.0957,
      "step": 11198
    },
    {
      "epoch": 0.8072514957110934,
      "grad_norm": 0.12848975259086523,
      "learning_rate": 1.8862915992422513e-05,
      "loss": 0.9878,
      "step": 11199
    },
    {
      "epoch": 0.8073235781734304,
      "grad_norm": 0.1295910710690706,
      "learning_rate": 1.8849271052935636e-05,
      "loss": 0.9979,
      "step": 11200
    },
    {
      "epoch": 0.8073956606357673,
      "grad_norm": 0.12951000632552814,
      "learning_rate": 1.8835630536889437e-05,
      "loss": 1.0721,
      "step": 11201
    },
    {
      "epoch": 0.8074677430981042,
      "grad_norm": 0.11572290257982654,
      "learning_rate": 1.8821994445027503e-05,
      "loss": 1.0608,
      "step": 11202
    },
    {
      "epoch": 0.8075398255604411,
      "grad_norm": 0.13880897787618052,
      "learning_rate": 1.8808362778093092e-05,
      "loss": 1.0003,
      "step": 11203
    },
    {
      "epoch": 0.807611908022778,
      "grad_norm": 0.12364775963788849,
      "learning_rate": 1.879473553682931e-05,
      "loss": 0.9901,
      "step": 11204
    },
    {
      "epoch": 0.807683990485115,
      "grad_norm": 0.12423937587432785,
      "learning_rate": 1.87811127219789e-05,
      "loss": 1.0344,
      "step": 11205
    },
    {
      "epoch": 0.8077560729474519,
      "grad_norm": 0.10879348932985752,
      "learning_rate": 1.8767494334284453e-05,
      "loss": 1.0391,
      "step": 11206
    },
    {
      "epoch": 0.8078281554097888,
      "grad_norm": 0.3469973298559053,
      "learning_rate": 1.8753880374488274e-05,
      "loss": 1.054,
      "step": 11207
    },
    {
      "epoch": 0.8079002378721257,
      "grad_norm": 0.5215237898771932,
      "learning_rate": 1.874027084333251e-05,
      "loss": 1.027,
      "step": 11208
    },
    {
      "epoch": 0.8079723203344626,
      "grad_norm": 0.13970878632055667,
      "learning_rate": 1.8726665741558923e-05,
      "loss": 1.0281,
      "step": 11209
    },
    {
      "epoch": 0.8080444027967996,
      "grad_norm": 0.1243228553443143,
      "learning_rate": 1.8713065069909152e-05,
      "loss": 1.061,
      "step": 11210
    },
    {
      "epoch": 0.8081164852591365,
      "grad_norm": 0.12014582142717832,
      "learning_rate": 1.8699468829124545e-05,
      "loss": 0.9705,
      "step": 11211
    },
    {
      "epoch": 0.8081885677214734,
      "grad_norm": 0.11487424039904087,
      "learning_rate": 1.8685877019946217e-05,
      "loss": 1.05,
      "step": 11212
    },
    {
      "epoch": 0.8082606501838103,
      "grad_norm": 0.12187641255075898,
      "learning_rate": 1.8672289643115058e-05,
      "loss": 1.0105,
      "step": 11213
    },
    {
      "epoch": 0.8083327326461472,
      "grad_norm": 0.11492573659016826,
      "learning_rate": 1.8658706699371687e-05,
      "loss": 1.043,
      "step": 11214
    },
    {
      "epoch": 0.808404815108484,
      "grad_norm": 0.12942431223670997,
      "learning_rate": 1.8645128189456505e-05,
      "loss": 1.0452,
      "step": 11215
    },
    {
      "epoch": 0.808476897570821,
      "grad_norm": 0.15686179094546923,
      "learning_rate": 1.863155411410965e-05,
      "loss": 1.0498,
      "step": 11216
    },
    {
      "epoch": 0.8085489800331579,
      "grad_norm": 0.10906525252352635,
      "learning_rate": 1.861798447407105e-05,
      "loss": 0.9295,
      "step": 11217
    },
    {
      "epoch": 0.8086210624954948,
      "grad_norm": 0.12539488791922404,
      "learning_rate": 1.8604419270080355e-05,
      "loss": 1.063,
      "step": 11218
    },
    {
      "epoch": 0.8086931449578317,
      "grad_norm": 0.12270945777950093,
      "learning_rate": 1.8590858502876983e-05,
      "loss": 0.9658,
      "step": 11219
    },
    {
      "epoch": 0.8087652274201687,
      "grad_norm": 0.14081099231516042,
      "learning_rate": 1.857730217320015e-05,
      "loss": 1.0094,
      "step": 11220
    },
    {
      "epoch": 0.8088373098825056,
      "grad_norm": 0.14520473056862934,
      "learning_rate": 1.8563750281788763e-05,
      "loss": 0.9928,
      "step": 11221
    },
    {
      "epoch": 0.8089093923448425,
      "grad_norm": 0.13218178194891755,
      "learning_rate": 1.8550202829381546e-05,
      "loss": 1.0105,
      "step": 11222
    },
    {
      "epoch": 0.8089814748071794,
      "grad_norm": 0.12012606378108989,
      "learning_rate": 1.853665981671694e-05,
      "loss": 1.0132,
      "step": 11223
    },
    {
      "epoch": 0.8090535572695163,
      "grad_norm": 0.18361929104537794,
      "learning_rate": 1.852312124453316e-05,
      "loss": 1.0163,
      "step": 11224
    },
    {
      "epoch": 0.8091256397318533,
      "grad_norm": 0.12268227373234886,
      "learning_rate": 1.8509587113568193e-05,
      "loss": 0.9993,
      "step": 11225
    },
    {
      "epoch": 0.8091977221941902,
      "grad_norm": 0.1285467624686648,
      "learning_rate": 1.8496057424559764e-05,
      "loss": 0.9721,
      "step": 11226
    },
    {
      "epoch": 0.8092698046565271,
      "grad_norm": 0.13062815187987772,
      "learning_rate": 1.848253217824536e-05,
      "loss": 1.0391,
      "step": 11227
    },
    {
      "epoch": 0.809341887118864,
      "grad_norm": 0.1201633649102055,
      "learning_rate": 1.8469011375362223e-05,
      "loss": 0.9276,
      "step": 11228
    },
    {
      "epoch": 0.8094139695812009,
      "grad_norm": 0.12089100666925527,
      "learning_rate": 1.845549501664736e-05,
      "loss": 0.9731,
      "step": 11229
    },
    {
      "epoch": 0.8094860520435379,
      "grad_norm": 0.12927562304080956,
      "learning_rate": 1.8441983102837546e-05,
      "loss": 1.0545,
      "step": 11230
    },
    {
      "epoch": 0.8095581345058748,
      "grad_norm": 0.12869786126780414,
      "learning_rate": 1.8428475634669285e-05,
      "loss": 0.9852,
      "step": 11231
    },
    {
      "epoch": 0.8096302169682116,
      "grad_norm": 0.16796858040724463,
      "learning_rate": 1.8414972612878867e-05,
      "loss": 1.0762,
      "step": 11232
    },
    {
      "epoch": 0.8097022994305485,
      "grad_norm": 0.13056994027513497,
      "learning_rate": 1.8401474038202337e-05,
      "loss": 1.0284,
      "step": 11233
    },
    {
      "epoch": 0.8097743818928854,
      "grad_norm": 0.11595010359052617,
      "learning_rate": 1.838797991137543e-05,
      "loss": 1.0117,
      "step": 11234
    },
    {
      "epoch": 0.8098464643552223,
      "grad_norm": 0.16543277157154823,
      "learning_rate": 1.8374490233133768e-05,
      "loss": 0.9874,
      "step": 11235
    },
    {
      "epoch": 0.8099185468175593,
      "grad_norm": 0.12044658854535538,
      "learning_rate": 1.8361005004212626e-05,
      "loss": 1.0906,
      "step": 11236
    },
    {
      "epoch": 0.8099906292798962,
      "grad_norm": 0.11780423436922492,
      "learning_rate": 1.834752422534711e-05,
      "loss": 1.0898,
      "step": 11237
    },
    {
      "epoch": 0.8100627117422331,
      "grad_norm": 0.11093044866607381,
      "learning_rate": 1.8334047897271945e-05,
      "loss": 1.1408,
      "step": 11238
    },
    {
      "epoch": 0.81013479420457,
      "grad_norm": 0.12390229693943415,
      "learning_rate": 1.8320576020721804e-05,
      "loss": 0.9209,
      "step": 11239
    },
    {
      "epoch": 0.8102068766669069,
      "grad_norm": 0.11754775566014566,
      "learning_rate": 1.8307108596431e-05,
      "loss": 1.0443,
      "step": 11240
    },
    {
      "epoch": 0.8102789591292439,
      "grad_norm": 0.13412815408304324,
      "learning_rate": 1.8293645625133636e-05,
      "loss": 1.0236,
      "step": 11241
    },
    {
      "epoch": 0.8103510415915808,
      "grad_norm": 0.13892393498225755,
      "learning_rate": 1.828018710756354e-05,
      "loss": 1.0094,
      "step": 11242
    },
    {
      "epoch": 0.8104231240539177,
      "grad_norm": 0.11561073985044769,
      "learning_rate": 1.8266733044454298e-05,
      "loss": 1.063,
      "step": 11243
    },
    {
      "epoch": 0.8104952065162546,
      "grad_norm": 0.15536354092524388,
      "learning_rate": 1.8253283436539338e-05,
      "loss": 0.9381,
      "step": 11244
    },
    {
      "epoch": 0.8105672889785915,
      "grad_norm": 0.13222464628082095,
      "learning_rate": 1.823983828455178e-05,
      "loss": 0.99,
      "step": 11245
    },
    {
      "epoch": 0.8106393714409285,
      "grad_norm": 0.24178983777429075,
      "learning_rate": 1.8226397589224464e-05,
      "loss": 1.0122,
      "step": 11246
    },
    {
      "epoch": 0.8107114539032654,
      "grad_norm": 0.1418389639355847,
      "learning_rate": 1.8212961351290025e-05,
      "loss": 0.9515,
      "step": 11247
    },
    {
      "epoch": 0.8107835363656023,
      "grad_norm": 0.15297576321874984,
      "learning_rate": 1.81995295714809e-05,
      "loss": 1.0094,
      "step": 11248
    },
    {
      "epoch": 0.8108556188279392,
      "grad_norm": 0.1161698800034809,
      "learning_rate": 1.8186102250529246e-05,
      "loss": 1.0088,
      "step": 11249
    },
    {
      "epoch": 0.810927701290276,
      "grad_norm": 0.3017693310097164,
      "learning_rate": 1.8172679389166925e-05,
      "loss": 1.0012,
      "step": 11250
    },
    {
      "epoch": 0.810999783752613,
      "grad_norm": 0.12380809482714752,
      "learning_rate": 1.815926098812559e-05,
      "loss": 1.0102,
      "step": 11251
    },
    {
      "epoch": 0.8110718662149499,
      "grad_norm": 0.12408065252373951,
      "learning_rate": 1.8145847048136734e-05,
      "loss": 1.0224,
      "step": 11252
    },
    {
      "epoch": 0.8111439486772868,
      "grad_norm": 0.1596183805777013,
      "learning_rate": 1.8132437569931527e-05,
      "loss": 1.036,
      "step": 11253
    },
    {
      "epoch": 0.8112160311396237,
      "grad_norm": 0.11625431097474571,
      "learning_rate": 1.8119032554240855e-05,
      "loss": 1.0758,
      "step": 11254
    },
    {
      "epoch": 0.8112881136019606,
      "grad_norm": 0.15222666566993073,
      "learning_rate": 1.8105632001795435e-05,
      "loss": 1.0078,
      "step": 11255
    },
    {
      "epoch": 0.8113601960642975,
      "grad_norm": 0.12218102620801126,
      "learning_rate": 1.8092235913325695e-05,
      "loss": 1.1225,
      "step": 11256
    },
    {
      "epoch": 0.8114322785266345,
      "grad_norm": 0.15945677099562539,
      "learning_rate": 1.8078844289561914e-05,
      "loss": 1.0007,
      "step": 11257
    },
    {
      "epoch": 0.8115043609889714,
      "grad_norm": 0.12170804451273509,
      "learning_rate": 1.8065457131233987e-05,
      "loss": 0.9637,
      "step": 11258
    },
    {
      "epoch": 0.8115764434513083,
      "grad_norm": 0.11653156867255626,
      "learning_rate": 1.8052074439071653e-05,
      "loss": 1.0536,
      "step": 11259
    },
    {
      "epoch": 0.8116485259136452,
      "grad_norm": 0.11562174117095375,
      "learning_rate": 1.8038696213804363e-05,
      "loss": 1.0209,
      "step": 11260
    },
    {
      "epoch": 0.8117206083759821,
      "grad_norm": 0.11478444303120193,
      "learning_rate": 1.802532245616142e-05,
      "loss": 1.0002,
      "step": 11261
    },
    {
      "epoch": 0.8117926908383191,
      "grad_norm": 0.13859398678272014,
      "learning_rate": 1.8011953166871753e-05,
      "loss": 1.0536,
      "step": 11262
    },
    {
      "epoch": 0.811864773300656,
      "grad_norm": 0.11819926032556251,
      "learning_rate": 1.7998588346664115e-05,
      "loss": 1.0095,
      "step": 11263
    },
    {
      "epoch": 0.8119368557629929,
      "grad_norm": 0.11157614537627113,
      "learning_rate": 1.798522799626702e-05,
      "loss": 0.9483,
      "step": 11264
    },
    {
      "epoch": 0.8120089382253298,
      "grad_norm": 0.1254451695974743,
      "learning_rate": 1.7971872116408727e-05,
      "loss": 0.9969,
      "step": 11265
    },
    {
      "epoch": 0.8120810206876667,
      "grad_norm": 0.1094187378418593,
      "learning_rate": 1.795852070781725e-05,
      "loss": 1.0248,
      "step": 11266
    },
    {
      "epoch": 0.8121531031500036,
      "grad_norm": 0.1255860279487947,
      "learning_rate": 1.794517377122036e-05,
      "loss": 1.035,
      "step": 11267
    },
    {
      "epoch": 0.8122251856123405,
      "grad_norm": 0.11453402218826694,
      "learning_rate": 1.793183130734558e-05,
      "loss": 1.0636,
      "step": 11268
    },
    {
      "epoch": 0.8122972680746774,
      "grad_norm": 0.11419692425624962,
      "learning_rate": 1.791849331692019e-05,
      "loss": 1.0571,
      "step": 11269
    },
    {
      "epoch": 0.8123693505370143,
      "grad_norm": 0.1137283524621901,
      "learning_rate": 1.7905159800671246e-05,
      "loss": 1.029,
      "step": 11270
    },
    {
      "epoch": 0.8124414329993512,
      "grad_norm": 0.19266147426921704,
      "learning_rate": 1.7891830759325534e-05,
      "loss": 1.0743,
      "step": 11271
    },
    {
      "epoch": 0.8125135154616882,
      "grad_norm": 0.1235055536766026,
      "learning_rate": 1.7878506193609602e-05,
      "loss": 1.0397,
      "step": 11272
    },
    {
      "epoch": 0.8125855979240251,
      "grad_norm": 0.13823321790197327,
      "learning_rate": 1.7865186104249764e-05,
      "loss": 1.0098,
      "step": 11273
    },
    {
      "epoch": 0.812657680386362,
      "grad_norm": 0.11229670421869122,
      "learning_rate": 1.785187049197208e-05,
      "loss": 1.0184,
      "step": 11274
    },
    {
      "epoch": 0.8127297628486989,
      "grad_norm": 0.10252133029712084,
      "learning_rate": 1.783855935750238e-05,
      "loss": 0.916,
      "step": 11275
    },
    {
      "epoch": 0.8128018453110358,
      "grad_norm": 0.1146380316651623,
      "learning_rate": 1.782525270156622e-05,
      "loss": 1.0415,
      "step": 11276
    },
    {
      "epoch": 0.8128739277733728,
      "grad_norm": 0.12500492882597547,
      "learning_rate": 1.7811950524888955e-05,
      "loss": 1.0034,
      "step": 11277
    },
    {
      "epoch": 0.8129460102357097,
      "grad_norm": 0.11571112434778236,
      "learning_rate": 1.779865282819566e-05,
      "loss": 1.1072,
      "step": 11278
    },
    {
      "epoch": 0.8130180926980466,
      "grad_norm": 0.13629421814735385,
      "learning_rate": 1.778535961221117e-05,
      "loss": 1.0519,
      "step": 11279
    },
    {
      "epoch": 0.8130901751603835,
      "grad_norm": 0.13345400970403914,
      "learning_rate": 1.7772070877660107e-05,
      "loss": 1.0535,
      "step": 11280
    },
    {
      "epoch": 0.8131622576227204,
      "grad_norm": 0.16364502318379093,
      "learning_rate": 1.775878662526681e-05,
      "loss": 1.0822,
      "step": 11281
    },
    {
      "epoch": 0.8132343400850574,
      "grad_norm": 0.13582595098387887,
      "learning_rate": 1.774550685575539e-05,
      "loss": 1.0341,
      "step": 11282
    },
    {
      "epoch": 0.8133064225473943,
      "grad_norm": 0.20208366703774278,
      "learning_rate": 1.773223156984971e-05,
      "loss": 1.0099,
      "step": 11283
    },
    {
      "epoch": 0.8133785050097311,
      "grad_norm": 0.12524450416711858,
      "learning_rate": 1.7718960768273406e-05,
      "loss": 1.0535,
      "step": 11284
    },
    {
      "epoch": 0.813450587472068,
      "grad_norm": 0.10851477504683439,
      "learning_rate": 1.7705694451749856e-05,
      "loss": 1.0255,
      "step": 11285
    },
    {
      "epoch": 0.8135226699344049,
      "grad_norm": 0.1132008459691898,
      "learning_rate": 1.7692432621002196e-05,
      "loss": 1.0252,
      "step": 11286
    },
    {
      "epoch": 0.8135947523967418,
      "grad_norm": 0.12221511676487501,
      "learning_rate": 1.7679175276753257e-05,
      "loss": 1.0895,
      "step": 11287
    },
    {
      "epoch": 0.8136668348590788,
      "grad_norm": 0.12044940732839267,
      "learning_rate": 1.7665922419725757e-05,
      "loss": 0.9895,
      "step": 11288
    },
    {
      "epoch": 0.8137389173214157,
      "grad_norm": 0.13756107744932516,
      "learning_rate": 1.765267405064206e-05,
      "loss": 1.0406,
      "step": 11289
    },
    {
      "epoch": 0.8138109997837526,
      "grad_norm": 0.12148793756774766,
      "learning_rate": 1.7639430170224357e-05,
      "loss": 1.0038,
      "step": 11290
    },
    {
      "epoch": 0.8138830822460895,
      "grad_norm": 0.11165318863312669,
      "learning_rate": 1.7626190779194475e-05,
      "loss": 1.0373,
      "step": 11291
    },
    {
      "epoch": 0.8139551647084264,
      "grad_norm": 0.1298528155092584,
      "learning_rate": 1.761295587827416e-05,
      "loss": 1.0426,
      "step": 11292
    },
    {
      "epoch": 0.8140272471707634,
      "grad_norm": 0.1284471264645122,
      "learning_rate": 1.7599725468184802e-05,
      "loss": 1.0597,
      "step": 11293
    },
    {
      "epoch": 0.8140993296331003,
      "grad_norm": 0.12432473614345103,
      "learning_rate": 1.7586499549647607e-05,
      "loss": 1.126,
      "step": 11294
    },
    {
      "epoch": 0.8141714120954372,
      "grad_norm": 0.13082589554471474,
      "learning_rate": 1.757327812338343e-05,
      "loss": 1.0424,
      "step": 11295
    },
    {
      "epoch": 0.8142434945577741,
      "grad_norm": 0.11183045501349949,
      "learning_rate": 1.756006119011303e-05,
      "loss": 1.0458,
      "step": 11296
    },
    {
      "epoch": 0.814315577020111,
      "grad_norm": 0.11945741996751665,
      "learning_rate": 1.754684875055682e-05,
      "loss": 1.047,
      "step": 11297
    },
    {
      "epoch": 0.814387659482448,
      "grad_norm": 0.12569460884315528,
      "learning_rate": 1.7533640805435038e-05,
      "loss": 1.0159,
      "step": 11298
    },
    {
      "epoch": 0.8144597419447849,
      "grad_norm": 0.11940709267057316,
      "learning_rate": 1.7520437355467555e-05,
      "loss": 1.0277,
      "step": 11299
    },
    {
      "epoch": 0.8145318244071218,
      "grad_norm": 0.11533291174247562,
      "learning_rate": 1.7507238401374105e-05,
      "loss": 1.0466,
      "step": 11300
    },
    {
      "epoch": 0.8146039068694587,
      "grad_norm": 0.13257722937015803,
      "learning_rate": 1.7494043943874183e-05,
      "loss": 0.959,
      "step": 11301
    },
    {
      "epoch": 0.8146759893317955,
      "grad_norm": 0.11702823024772178,
      "learning_rate": 1.7480853983687018e-05,
      "loss": 1.0061,
      "step": 11302
    },
    {
      "epoch": 0.8147480717941324,
      "grad_norm": 0.16218875796445678,
      "learning_rate": 1.7467668521531523e-05,
      "loss": 1.1327,
      "step": 11303
    },
    {
      "epoch": 0.8148201542564694,
      "grad_norm": 0.14477318895701005,
      "learning_rate": 1.745448755812641e-05,
      "loss": 1.0089,
      "step": 11304
    },
    {
      "epoch": 0.8148922367188063,
      "grad_norm": 0.16161598818021858,
      "learning_rate": 1.744131109419024e-05,
      "loss": 1.0441,
      "step": 11305
    },
    {
      "epoch": 0.8149643191811432,
      "grad_norm": 0.11368596194499542,
      "learning_rate": 1.7428139130441224e-05,
      "loss": 1.0517,
      "step": 11306
    },
    {
      "epoch": 0.8150364016434801,
      "grad_norm": 0.16838393212353914,
      "learning_rate": 1.7414971667597314e-05,
      "loss": 0.9356,
      "step": 11307
    },
    {
      "epoch": 0.815108484105817,
      "grad_norm": 0.11858559416028785,
      "learning_rate": 1.7401808706376267e-05,
      "loss": 1.0463,
      "step": 11308
    },
    {
      "epoch": 0.815180566568154,
      "grad_norm": 0.11756502963481923,
      "learning_rate": 1.7388650247495565e-05,
      "loss": 1.0372,
      "step": 11309
    },
    {
      "epoch": 0.8152526490304909,
      "grad_norm": 0.11884382094542846,
      "learning_rate": 1.7375496291672545e-05,
      "loss": 1.058,
      "step": 11310
    },
    {
      "epoch": 0.8153247314928278,
      "grad_norm": 0.10947097776000557,
      "learning_rate": 1.7362346839624123e-05,
      "loss": 0.9861,
      "step": 11311
    },
    {
      "epoch": 0.8153968139551647,
      "grad_norm": 0.12484053577276999,
      "learning_rate": 1.7349201892067103e-05,
      "loss": 1.0449,
      "step": 11312
    },
    {
      "epoch": 0.8154688964175016,
      "grad_norm": 0.13790328014535144,
      "learning_rate": 1.7336061449717966e-05,
      "loss": 1.0543,
      "step": 11313
    },
    {
      "epoch": 0.8155409788798386,
      "grad_norm": 0.22913274702450565,
      "learning_rate": 1.7322925513293064e-05,
      "loss": 1.0569,
      "step": 11314
    },
    {
      "epoch": 0.8156130613421755,
      "grad_norm": 0.12147038456324347,
      "learning_rate": 1.7309794083508346e-05,
      "loss": 0.991,
      "step": 11315
    },
    {
      "epoch": 0.8156851438045124,
      "grad_norm": 0.3568129734972585,
      "learning_rate": 1.7296667161079615e-05,
      "loss": 1.0877,
      "step": 11316
    },
    {
      "epoch": 0.8157572262668493,
      "grad_norm": 0.11367513378564843,
      "learning_rate": 1.728354474672239e-05,
      "loss": 1.073,
      "step": 11317
    },
    {
      "epoch": 0.8158293087291862,
      "grad_norm": 0.21640784967782684,
      "learning_rate": 1.7270426841152032e-05,
      "loss": 1.0963,
      "step": 11318
    },
    {
      "epoch": 0.815901391191523,
      "grad_norm": 0.11709377108108518,
      "learning_rate": 1.72573134450835e-05,
      "loss": 0.9969,
      "step": 11319
    },
    {
      "epoch": 0.81597347365386,
      "grad_norm": 0.12871142108047945,
      "learning_rate": 1.7244204559231624e-05,
      "loss": 1.0819,
      "step": 11320
    },
    {
      "epoch": 0.8160455561161969,
      "grad_norm": 0.34479868889540694,
      "learning_rate": 1.7231100184310956e-05,
      "loss": 1.0351,
      "step": 11321
    },
    {
      "epoch": 0.8161176385785338,
      "grad_norm": 0.1224410182457978,
      "learning_rate": 1.7218000321035797e-05,
      "loss": 0.9702,
      "step": 11322
    },
    {
      "epoch": 0.8161897210408707,
      "grad_norm": 0.13035746655712221,
      "learning_rate": 1.7204904970120207e-05,
      "loss": 1.0041,
      "step": 11323
    },
    {
      "epoch": 0.8162618035032077,
      "grad_norm": 0.13230352903925086,
      "learning_rate": 1.7191814132278007e-05,
      "loss": 1.0225,
      "step": 11324
    },
    {
      "epoch": 0.8163338859655446,
      "grad_norm": 0.13620890769121702,
      "learning_rate": 1.717872780822275e-05,
      "loss": 0.9426,
      "step": 11325
    },
    {
      "epoch": 0.8164059684278815,
      "grad_norm": 0.12774019509449056,
      "learning_rate": 1.716564599866778e-05,
      "loss": 1.0631,
      "step": 11326
    },
    {
      "epoch": 0.8164780508902184,
      "grad_norm": 0.1301797451040631,
      "learning_rate": 1.7152568704326156e-05,
      "loss": 1.0335,
      "step": 11327
    },
    {
      "epoch": 0.8165501333525553,
      "grad_norm": 0.1292131220367316,
      "learning_rate": 1.713949592591071e-05,
      "loss": 1.0322,
      "step": 11328
    },
    {
      "epoch": 0.8166222158148922,
      "grad_norm": 0.15951148865321416,
      "learning_rate": 1.7126427664134027e-05,
      "loss": 1.0013,
      "step": 11329
    },
    {
      "epoch": 0.8166942982772292,
      "grad_norm": 0.11697321846107205,
      "learning_rate": 1.711336391970845e-05,
      "loss": 1.0698,
      "step": 11330
    },
    {
      "epoch": 0.8167663807395661,
      "grad_norm": 0.15214331036691284,
      "learning_rate": 1.710030469334606e-05,
      "loss": 1.1154,
      "step": 11331
    },
    {
      "epoch": 0.816838463201903,
      "grad_norm": 0.17534233827641127,
      "learning_rate": 1.7087249985758715e-05,
      "loss": 0.9836,
      "step": 11332
    },
    {
      "epoch": 0.8169105456642399,
      "grad_norm": 0.11435498951627432,
      "learning_rate": 1.7074199797657984e-05,
      "loss": 1.0158,
      "step": 11333
    },
    {
      "epoch": 0.8169826281265768,
      "grad_norm": 0.1507584170425034,
      "learning_rate": 1.7061154129755252e-05,
      "loss": 1.0669,
      "step": 11334
    },
    {
      "epoch": 0.8170547105889138,
      "grad_norm": 0.36659569728399105,
      "learning_rate": 1.7048112982761622e-05,
      "loss": 1.0905,
      "step": 11335
    },
    {
      "epoch": 0.8171267930512506,
      "grad_norm": 0.1299866835495928,
      "learning_rate": 1.7035076357387924e-05,
      "loss": 0.9812,
      "step": 11336
    },
    {
      "epoch": 0.8171988755135875,
      "grad_norm": 0.10898354825455536,
      "learning_rate": 1.7022044254344805e-05,
      "loss": 1.0466,
      "step": 11337
    },
    {
      "epoch": 0.8172709579759244,
      "grad_norm": 0.14239928615632388,
      "learning_rate": 1.7009016674342614e-05,
      "loss": 1.0739,
      "step": 11338
    },
    {
      "epoch": 0.8173430404382613,
      "grad_norm": 0.12901029864820837,
      "learning_rate": 1.6995993618091467e-05,
      "loss": 1.0593,
      "step": 11339
    },
    {
      "epoch": 0.8174151229005983,
      "grad_norm": 0.1324967802515352,
      "learning_rate": 1.6982975086301244e-05,
      "loss": 0.9801,
      "step": 11340
    },
    {
      "epoch": 0.8174872053629352,
      "grad_norm": 0.11412813583555445,
      "learning_rate": 1.6969961079681574e-05,
      "loss": 1.1,
      "step": 11341
    },
    {
      "epoch": 0.8175592878252721,
      "grad_norm": 0.12595459389990965,
      "learning_rate": 1.6956951598941828e-05,
      "loss": 1.019,
      "step": 11342
    },
    {
      "epoch": 0.817631370287609,
      "grad_norm": 0.11792916987043646,
      "learning_rate": 1.6943946644791164e-05,
      "loss": 0.9924,
      "step": 11343
    },
    {
      "epoch": 0.8177034527499459,
      "grad_norm": 0.12443114229662336,
      "learning_rate": 1.6930946217938416e-05,
      "loss": 1.0501,
      "step": 11344
    },
    {
      "epoch": 0.8177755352122829,
      "grad_norm": 0.10957071717141287,
      "learning_rate": 1.691795031909227e-05,
      "loss": 0.9885,
      "step": 11345
    },
    {
      "epoch": 0.8178476176746198,
      "grad_norm": 0.13121724919621283,
      "learning_rate": 1.6904958948961104e-05,
      "loss": 1.0389,
      "step": 11346
    },
    {
      "epoch": 0.8179197001369567,
      "grad_norm": 0.14221405388755134,
      "learning_rate": 1.6891972108253095e-05,
      "loss": 1.0436,
      "step": 11347
    },
    {
      "epoch": 0.8179917825992936,
      "grad_norm": 0.1476726162050856,
      "learning_rate": 1.687898979767606e-05,
      "loss": 0.9711,
      "step": 11348
    },
    {
      "epoch": 0.8180638650616305,
      "grad_norm": 0.17106852123297037,
      "learning_rate": 1.6866012017937715e-05,
      "loss": 1.0593,
      "step": 11349
    },
    {
      "epoch": 0.8181359475239675,
      "grad_norm": 0.13443847364092149,
      "learning_rate": 1.6853038769745467e-05,
      "loss": 1.0708,
      "step": 11350
    },
    {
      "epoch": 0.8182080299863044,
      "grad_norm": 0.13260197506879234,
      "learning_rate": 1.684007005380649e-05,
      "loss": 0.9646,
      "step": 11351
    },
    {
      "epoch": 0.8182801124486413,
      "grad_norm": 0.11691833301967751,
      "learning_rate": 1.682710587082763e-05,
      "loss": 1.0809,
      "step": 11352
    },
    {
      "epoch": 0.8183521949109781,
      "grad_norm": 0.13095113007044573,
      "learning_rate": 1.681414622151556e-05,
      "loss": 0.9944,
      "step": 11353
    },
    {
      "epoch": 0.818424277373315,
      "grad_norm": 0.24295659384852009,
      "learning_rate": 1.6801191106576775e-05,
      "loss": 1.1278,
      "step": 11354
    },
    {
      "epoch": 0.8184963598356519,
      "grad_norm": 0.12248273514524717,
      "learning_rate": 1.6788240526717368e-05,
      "loss": 1.0137,
      "step": 11355
    },
    {
      "epoch": 0.8185684422979889,
      "grad_norm": 0.1093912946367584,
      "learning_rate": 1.677529448264329e-05,
      "loss": 1.089,
      "step": 11356
    },
    {
      "epoch": 0.8186405247603258,
      "grad_norm": 0.11650347446229731,
      "learning_rate": 1.676235297506018e-05,
      "loss": 1.0212,
      "step": 11357
    },
    {
      "epoch": 0.8187126072226627,
      "grad_norm": 0.17695418760927714,
      "learning_rate": 1.6749416004673558e-05,
      "loss": 1.0431,
      "step": 11358
    },
    {
      "epoch": 0.8187846896849996,
      "grad_norm": 0.1390442781917121,
      "learning_rate": 1.673648357218851e-05,
      "loss": 1.0654,
      "step": 11359
    },
    {
      "epoch": 0.8188567721473365,
      "grad_norm": 0.6304527846536788,
      "learning_rate": 1.672355567831001e-05,
      "loss": 1.0938,
      "step": 11360
    },
    {
      "epoch": 0.8189288546096735,
      "grad_norm": 0.1355105336468707,
      "learning_rate": 1.6710632323742738e-05,
      "loss": 1.0988,
      "step": 11361
    },
    {
      "epoch": 0.8190009370720104,
      "grad_norm": 0.12813515651294755,
      "learning_rate": 1.6697713509191125e-05,
      "loss": 0.9828,
      "step": 11362
    },
    {
      "epoch": 0.8190730195343473,
      "grad_norm": 0.12400545400397042,
      "learning_rate": 1.668479923535937e-05,
      "loss": 1.0615,
      "step": 11363
    },
    {
      "epoch": 0.8191451019966842,
      "grad_norm": 0.14907543013655777,
      "learning_rate": 1.6671889502951432e-05,
      "loss": 1.0473,
      "step": 11364
    },
    {
      "epoch": 0.8192171844590211,
      "grad_norm": 0.12280234186796676,
      "learning_rate": 1.6658984312670978e-05,
      "loss": 1.0861,
      "step": 11365
    },
    {
      "epoch": 0.8192892669213581,
      "grad_norm": 0.12477592128165181,
      "learning_rate": 1.664608366522147e-05,
      "loss": 1.0174,
      "step": 11366
    },
    {
      "epoch": 0.819361349383695,
      "grad_norm": 0.10774498694549992,
      "learning_rate": 1.6633187561306108e-05,
      "loss": 1.0207,
      "step": 11367
    },
    {
      "epoch": 0.8194334318460319,
      "grad_norm": 0.21084302698953028,
      "learning_rate": 1.6620296001627845e-05,
      "loss": 1.028,
      "step": 11368
    },
    {
      "epoch": 0.8195055143083688,
      "grad_norm": 0.10229617854762192,
      "learning_rate": 1.6607408986889384e-05,
      "loss": 0.9497,
      "step": 11369
    },
    {
      "epoch": 0.8195775967707057,
      "grad_norm": 0.11889864035211338,
      "learning_rate": 1.6594526517793185e-05,
      "loss": 1.0008,
      "step": 11370
    },
    {
      "epoch": 0.8196496792330425,
      "grad_norm": 0.11894372095179834,
      "learning_rate": 1.6581648595041454e-05,
      "loss": 1.0807,
      "step": 11371
    },
    {
      "epoch": 0.8197217616953795,
      "grad_norm": 0.12004367261102686,
      "learning_rate": 1.6568775219336164e-05,
      "loss": 0.977,
      "step": 11372
    },
    {
      "epoch": 0.8197938441577164,
      "grad_norm": 0.13370988931371736,
      "learning_rate": 1.6555906391379015e-05,
      "loss": 1.0686,
      "step": 11373
    },
    {
      "epoch": 0.8198659266200533,
      "grad_norm": 0.13820348304232843,
      "learning_rate": 1.6543042111871486e-05,
      "loss": 1.0961,
      "step": 11374
    },
    {
      "epoch": 0.8199380090823902,
      "grad_norm": 0.13279144167282073,
      "learning_rate": 1.653018238151478e-05,
      "loss": 0.9339,
      "step": 11375
    },
    {
      "epoch": 0.8200100915447271,
      "grad_norm": 0.10956110884643461,
      "learning_rate": 1.6517327201009893e-05,
      "loss": 1.0165,
      "step": 11376
    },
    {
      "epoch": 0.8200821740070641,
      "grad_norm": 0.12714496926442223,
      "learning_rate": 1.6504476571057516e-05,
      "loss": 1.0306,
      "step": 11377
    },
    {
      "epoch": 0.820154256469401,
      "grad_norm": 0.11425470982233486,
      "learning_rate": 1.6491630492358147e-05,
      "loss": 1.0305,
      "step": 11378
    },
    {
      "epoch": 0.8202263389317379,
      "grad_norm": 0.1171528248021307,
      "learning_rate": 1.6478788965611993e-05,
      "loss": 1.0375,
      "step": 11379
    },
    {
      "epoch": 0.8202984213940748,
      "grad_norm": 0.11598310577823338,
      "learning_rate": 1.6465951991519045e-05,
      "loss": 1.0255,
      "step": 11380
    },
    {
      "epoch": 0.8203705038564117,
      "grad_norm": 0.12404441122283157,
      "learning_rate": 1.6453119570779042e-05,
      "loss": 0.9988,
      "step": 11381
    },
    {
      "epoch": 0.8204425863187487,
      "grad_norm": 0.1338304697025758,
      "learning_rate": 1.644029170409145e-05,
      "loss": 0.973,
      "step": 11382
    },
    {
      "epoch": 0.8205146687810856,
      "grad_norm": 0.12075787850938885,
      "learning_rate": 1.642746839215552e-05,
      "loss": 0.9637,
      "step": 11383
    },
    {
      "epoch": 0.8205867512434225,
      "grad_norm": 0.16881455436429432,
      "learning_rate": 1.6414649635670188e-05,
      "loss": 1.0075,
      "step": 11384
    },
    {
      "epoch": 0.8206588337057594,
      "grad_norm": 0.21370829317169476,
      "learning_rate": 1.6401835435334245e-05,
      "loss": 1.0276,
      "step": 11385
    },
    {
      "epoch": 0.8207309161680963,
      "grad_norm": 0.16360591656193826,
      "learning_rate": 1.638902579184616e-05,
      "loss": 1.0731,
      "step": 11386
    },
    {
      "epoch": 0.8208029986304333,
      "grad_norm": 0.25386194226624953,
      "learning_rate": 1.63762207059042e-05,
      "loss": 1.0634,
      "step": 11387
    },
    {
      "epoch": 0.8208750810927701,
      "grad_norm": 0.11437664083880525,
      "learning_rate": 1.636342017820629e-05,
      "loss": 0.9887,
      "step": 11388
    },
    {
      "epoch": 0.820947163555107,
      "grad_norm": 0.14452889038200867,
      "learning_rate": 1.6350624209450226e-05,
      "loss": 1.0642,
      "step": 11389
    },
    {
      "epoch": 0.8210192460174439,
      "grad_norm": 0.13445391821403005,
      "learning_rate": 1.6337832800333496e-05,
      "loss": 1.0,
      "step": 11390
    },
    {
      "epoch": 0.8210913284797808,
      "grad_norm": 0.11861374260804171,
      "learning_rate": 1.6325045951553374e-05,
      "loss": 1.0492,
      "step": 11391
    },
    {
      "epoch": 0.8211634109421178,
      "grad_norm": 0.12591867716158037,
      "learning_rate": 1.631226366380677e-05,
      "loss": 1.0071,
      "step": 11392
    },
    {
      "epoch": 0.8212354934044547,
      "grad_norm": 0.12401214195547436,
      "learning_rate": 1.6299485937790502e-05,
      "loss": 1.0631,
      "step": 11393
    },
    {
      "epoch": 0.8213075758667916,
      "grad_norm": 0.11349705641523794,
      "learning_rate": 1.6286712774201074e-05,
      "loss": 0.986,
      "step": 11394
    },
    {
      "epoch": 0.8213796583291285,
      "grad_norm": 0.12811784634669018,
      "learning_rate": 1.6273944173734735e-05,
      "loss": 0.9854,
      "step": 11395
    },
    {
      "epoch": 0.8214517407914654,
      "grad_norm": 0.13712755845319152,
      "learning_rate": 1.626118013708745e-05,
      "loss": 1.0271,
      "step": 11396
    },
    {
      "epoch": 0.8215238232538024,
      "grad_norm": 0.11799868008103027,
      "learning_rate": 1.624842066495498e-05,
      "loss": 0.9959,
      "step": 11397
    },
    {
      "epoch": 0.8215959057161393,
      "grad_norm": 0.10738814666113865,
      "learning_rate": 1.623566575803286e-05,
      "loss": 0.9988,
      "step": 11398
    },
    {
      "epoch": 0.8216679881784762,
      "grad_norm": 0.1168420700117817,
      "learning_rate": 1.622291541701637e-05,
      "loss": 1.0407,
      "step": 11399
    },
    {
      "epoch": 0.8217400706408131,
      "grad_norm": 0.1483977513876191,
      "learning_rate": 1.621016964260045e-05,
      "loss": 1.0596,
      "step": 11400
    },
    {
      "epoch": 0.82181215310315,
      "grad_norm": 0.14163781682809273,
      "learning_rate": 1.6197428435479857e-05,
      "loss": 1.1026,
      "step": 11401
    },
    {
      "epoch": 0.821884235565487,
      "grad_norm": 0.114462574759524,
      "learning_rate": 1.6184691796349172e-05,
      "loss": 0.9907,
      "step": 11402
    },
    {
      "epoch": 0.8219563180278239,
      "grad_norm": 0.1262467003021503,
      "learning_rate": 1.617195972590263e-05,
      "loss": 1.034,
      "step": 11403
    },
    {
      "epoch": 0.8220284004901608,
      "grad_norm": 0.1234557603621009,
      "learning_rate": 1.615923222483421e-05,
      "loss": 1.0448,
      "step": 11404
    },
    {
      "epoch": 0.8221004829524976,
      "grad_norm": 0.13260446020456343,
      "learning_rate": 1.6146509293837698e-05,
      "loss": 1.0182,
      "step": 11405
    },
    {
      "epoch": 0.8221725654148345,
      "grad_norm": 0.16282700345448844,
      "learning_rate": 1.6133790933606573e-05,
      "loss": 0.9807,
      "step": 11406
    },
    {
      "epoch": 0.8222446478771714,
      "grad_norm": 0.13441077726933348,
      "learning_rate": 1.6121077144834185e-05,
      "loss": 1.0569,
      "step": 11407
    },
    {
      "epoch": 0.8223167303395084,
      "grad_norm": 0.1502509973354255,
      "learning_rate": 1.6108367928213476e-05,
      "loss": 0.9775,
      "step": 11408
    },
    {
      "epoch": 0.8223888128018453,
      "grad_norm": 0.11845447866352773,
      "learning_rate": 1.6095663284437222e-05,
      "loss": 0.9599,
      "step": 11409
    },
    {
      "epoch": 0.8224608952641822,
      "grad_norm": 0.11824864954114499,
      "learning_rate": 1.6082963214197923e-05,
      "loss": 1.0331,
      "step": 11410
    },
    {
      "epoch": 0.8225329777265191,
      "grad_norm": 0.1217866090158661,
      "learning_rate": 1.6070267718187927e-05,
      "loss": 1.0216,
      "step": 11411
    },
    {
      "epoch": 0.822605060188856,
      "grad_norm": 0.1499265289642477,
      "learning_rate": 1.605757679709917e-05,
      "loss": 1.0015,
      "step": 11412
    },
    {
      "epoch": 0.822677142651193,
      "grad_norm": 0.36388321353060615,
      "learning_rate": 1.604489045162344e-05,
      "loss": 1.0608,
      "step": 11413
    },
    {
      "epoch": 0.8227492251135299,
      "grad_norm": 0.11641726792381256,
      "learning_rate": 1.6032208682452254e-05,
      "loss": 0.9949,
      "step": 11414
    },
    {
      "epoch": 0.8228213075758668,
      "grad_norm": 0.6733971009601599,
      "learning_rate": 1.601953149027693e-05,
      "loss": 1.0129,
      "step": 11415
    },
    {
      "epoch": 0.8228933900382037,
      "grad_norm": 0.12082105251676818,
      "learning_rate": 1.6006858875788434e-05,
      "loss": 1.0528,
      "step": 11416
    },
    {
      "epoch": 0.8229654725005406,
      "grad_norm": 0.14238718208815146,
      "learning_rate": 1.5994190839677546e-05,
      "loss": 1.0434,
      "step": 11417
    },
    {
      "epoch": 0.8230375549628776,
      "grad_norm": 0.12945701310166424,
      "learning_rate": 1.5981527382634808e-05,
      "loss": 1.0544,
      "step": 11418
    },
    {
      "epoch": 0.8231096374252145,
      "grad_norm": 0.12653596223679703,
      "learning_rate": 1.5968868505350466e-05,
      "loss": 1.0565,
      "step": 11419
    },
    {
      "epoch": 0.8231817198875514,
      "grad_norm": 0.12862111055272996,
      "learning_rate": 1.5956214208514554e-05,
      "loss": 1.0561,
      "step": 11420
    },
    {
      "epoch": 0.8232538023498883,
      "grad_norm": 0.1499319582574653,
      "learning_rate": 1.5943564492816853e-05,
      "loss": 1.1272,
      "step": 11421
    },
    {
      "epoch": 0.8233258848122252,
      "grad_norm": 0.13563585402930609,
      "learning_rate": 1.5930919358946873e-05,
      "loss": 1.0416,
      "step": 11422
    },
    {
      "epoch": 0.823397967274562,
      "grad_norm": 0.11744786225467307,
      "learning_rate": 1.591827880759389e-05,
      "loss": 1.0812,
      "step": 11423
    },
    {
      "epoch": 0.823470049736899,
      "grad_norm": 0.12382220655680934,
      "learning_rate": 1.590564283944693e-05,
      "loss": 1.0453,
      "step": 11424
    },
    {
      "epoch": 0.8235421321992359,
      "grad_norm": 0.13217654591337316,
      "learning_rate": 1.5893011455194762e-05,
      "loss": 0.9526,
      "step": 11425
    },
    {
      "epoch": 0.8236142146615728,
      "grad_norm": 0.451977879476416,
      "learning_rate": 1.5880384655525914e-05,
      "loss": 1.0,
      "step": 11426
    },
    {
      "epoch": 0.8236862971239097,
      "grad_norm": 1.2737734161294334,
      "learning_rate": 1.5867762441128664e-05,
      "loss": 0.9748,
      "step": 11427
    },
    {
      "epoch": 0.8237583795862466,
      "grad_norm": 0.12663366427030417,
      "learning_rate": 1.5855144812691027e-05,
      "loss": 0.9873,
      "step": 11428
    },
    {
      "epoch": 0.8238304620485836,
      "grad_norm": 0.12880025523901406,
      "learning_rate": 1.584253177090077e-05,
      "loss": 1.1063,
      "step": 11429
    },
    {
      "epoch": 0.8239025445109205,
      "grad_norm": 0.2978279310208208,
      "learning_rate": 1.582992331644544e-05,
      "loss": 1.1124,
      "step": 11430
    },
    {
      "epoch": 0.8239746269732574,
      "grad_norm": 0.13270724670618197,
      "learning_rate": 1.581731945001228e-05,
      "loss": 1.0894,
      "step": 11431
    },
    {
      "epoch": 0.8240467094355943,
      "grad_norm": 0.11491623438912356,
      "learning_rate": 1.5804720172288344e-05,
      "loss": 1.0056,
      "step": 11432
    },
    {
      "epoch": 0.8241187918979312,
      "grad_norm": 0.13608144423098845,
      "learning_rate": 1.5792125483960385e-05,
      "loss": 1.02,
      "step": 11433
    },
    {
      "epoch": 0.8241908743602682,
      "grad_norm": 0.14857235205588867,
      "learning_rate": 1.5779535385714927e-05,
      "loss": 1.1024,
      "step": 11434
    },
    {
      "epoch": 0.8242629568226051,
      "grad_norm": 0.12740479391160095,
      "learning_rate": 1.5766949878238258e-05,
      "loss": 1.0083,
      "step": 11435
    },
    {
      "epoch": 0.824335039284942,
      "grad_norm": 0.13439990106200728,
      "learning_rate": 1.5754368962216393e-05,
      "loss": 1.0175,
      "step": 11436
    },
    {
      "epoch": 0.8244071217472789,
      "grad_norm": 0.12304434588405935,
      "learning_rate": 1.5741792638335095e-05,
      "loss": 1.0555,
      "step": 11437
    },
    {
      "epoch": 0.8244792042096158,
      "grad_norm": 0.168672877770791,
      "learning_rate": 1.5729220907279897e-05,
      "loss": 1.0081,
      "step": 11438
    },
    {
      "epoch": 0.8245512866719528,
      "grad_norm": 0.12186091317320034,
      "learning_rate": 1.5716653769736057e-05,
      "loss": 1.0278,
      "step": 11439
    },
    {
      "epoch": 0.8246233691342896,
      "grad_norm": 0.12526187604807856,
      "learning_rate": 1.570409122638864e-05,
      "loss": 1.0418,
      "step": 11440
    },
    {
      "epoch": 0.8246954515966265,
      "grad_norm": 0.13352237974557532,
      "learning_rate": 1.5691533277922343e-05,
      "loss": 0.9945,
      "step": 11441
    },
    {
      "epoch": 0.8247675340589634,
      "grad_norm": 0.17075892085141728,
      "learning_rate": 1.5678979925021743e-05,
      "loss": 1.057,
      "step": 11442
    },
    {
      "epoch": 0.8248396165213003,
      "grad_norm": 0.14446075858009577,
      "learning_rate": 1.5666431168371086e-05,
      "loss": 1.0291,
      "step": 11443
    },
    {
      "epoch": 0.8249116989836373,
      "grad_norm": 0.12399933029853122,
      "learning_rate": 1.5653887008654444e-05,
      "loss": 0.9867,
      "step": 11444
    },
    {
      "epoch": 0.8249837814459742,
      "grad_norm": 0.13789446078972445,
      "learning_rate": 1.5641347446555476e-05,
      "loss": 1.0327,
      "step": 11445
    },
    {
      "epoch": 0.8250558639083111,
      "grad_norm": 2.288863969140316,
      "learning_rate": 1.5628812482757803e-05,
      "loss": 1.0111,
      "step": 11446
    },
    {
      "epoch": 0.825127946370648,
      "grad_norm": 0.12034656963247792,
      "learning_rate": 1.5616282117944648e-05,
      "loss": 1.0887,
      "step": 11447
    },
    {
      "epoch": 0.8252000288329849,
      "grad_norm": 0.10897432231864738,
      "learning_rate": 1.560375635279906e-05,
      "loss": 1.0499,
      "step": 11448
    },
    {
      "epoch": 0.8252721112953219,
      "grad_norm": 0.10463532870980995,
      "learning_rate": 1.559123518800376e-05,
      "loss": 0.9863,
      "step": 11449
    },
    {
      "epoch": 0.8253441937576588,
      "grad_norm": 0.13926085998069024,
      "learning_rate": 1.5578718624241262e-05,
      "loss": 1.0591,
      "step": 11450
    },
    {
      "epoch": 0.8254162762199957,
      "grad_norm": 0.11599933122879343,
      "learning_rate": 1.556620666219387e-05,
      "loss": 1.0579,
      "step": 11451
    },
    {
      "epoch": 0.8254883586823326,
      "grad_norm": 0.13140765568354512,
      "learning_rate": 1.555369930254361e-05,
      "loss": 0.9646,
      "step": 11452
    },
    {
      "epoch": 0.8255604411446695,
      "grad_norm": 0.13150065997932606,
      "learning_rate": 1.55411965459722e-05,
      "loss": 1.0228,
      "step": 11453
    },
    {
      "epoch": 0.8256325236070065,
      "grad_norm": 0.8170939985811412,
      "learning_rate": 1.5528698393161134e-05,
      "loss": 1.009,
      "step": 11454
    },
    {
      "epoch": 0.8257046060693434,
      "grad_norm": 0.12284355368756732,
      "learning_rate": 1.551620484479174e-05,
      "loss": 1.0662,
      "step": 11455
    },
    {
      "epoch": 0.8257766885316803,
      "grad_norm": 0.1227788964010644,
      "learning_rate": 1.5503715901545013e-05,
      "loss": 0.9972,
      "step": 11456
    },
    {
      "epoch": 0.8258487709940171,
      "grad_norm": 0.12917860159480443,
      "learning_rate": 1.5491231564101672e-05,
      "loss": 1.1146,
      "step": 11457
    },
    {
      "epoch": 0.825920853456354,
      "grad_norm": 2.516384934530829,
      "learning_rate": 1.5478751833142224e-05,
      "loss": 1.0106,
      "step": 11458
    },
    {
      "epoch": 0.8259929359186909,
      "grad_norm": 0.12199906337056818,
      "learning_rate": 1.546627670934697e-05,
      "loss": 1.0558,
      "step": 11459
    },
    {
      "epoch": 0.8260650183810279,
      "grad_norm": 0.11883160332052238,
      "learning_rate": 1.5453806193395925e-05,
      "loss": 0.9862,
      "step": 11460
    },
    {
      "epoch": 0.8261371008433648,
      "grad_norm": 0.11827154598415285,
      "learning_rate": 1.5441340285968786e-05,
      "loss": 1.0227,
      "step": 11461
    },
    {
      "epoch": 0.8262091833057017,
      "grad_norm": 0.12315008431770451,
      "learning_rate": 1.5428878987745086e-05,
      "loss": 1.0481,
      "step": 11462
    },
    {
      "epoch": 0.8262812657680386,
      "grad_norm": 0.1314794651066684,
      "learning_rate": 1.5416422299404053e-05,
      "loss": 0.9939,
      "step": 11463
    },
    {
      "epoch": 0.8263533482303755,
      "grad_norm": 0.15575192348677702,
      "learning_rate": 1.5403970221624763e-05,
      "loss": 1.0229,
      "step": 11464
    },
    {
      "epoch": 0.8264254306927125,
      "grad_norm": 0.12518419207350795,
      "learning_rate": 1.5391522755085884e-05,
      "loss": 1.0714,
      "step": 11465
    },
    {
      "epoch": 0.8264975131550494,
      "grad_norm": 0.11831817568519822,
      "learning_rate": 1.5379079900465953e-05,
      "loss": 1.064,
      "step": 11466
    },
    {
      "epoch": 0.8265695956173863,
      "grad_norm": 0.12377158276900115,
      "learning_rate": 1.536664165844317e-05,
      "loss": 1.0073,
      "step": 11467
    },
    {
      "epoch": 0.8266416780797232,
      "grad_norm": 0.13225348086541439,
      "learning_rate": 1.5354208029695628e-05,
      "loss": 0.9825,
      "step": 11468
    },
    {
      "epoch": 0.8267137605420601,
      "grad_norm": 0.11894928619758427,
      "learning_rate": 1.5341779014900982e-05,
      "loss": 1.0742,
      "step": 11469
    },
    {
      "epoch": 0.8267858430043971,
      "grad_norm": 0.12651519625387941,
      "learning_rate": 1.5329354614736757e-05,
      "loss": 1.0477,
      "step": 11470
    },
    {
      "epoch": 0.826857925466734,
      "grad_norm": 0.12590694127172916,
      "learning_rate": 1.531693482988018e-05,
      "loss": 0.9689,
      "step": 11471
    },
    {
      "epoch": 0.8269300079290709,
      "grad_norm": 0.14798980571419218,
      "learning_rate": 1.5304519661008267e-05,
      "loss": 1.0892,
      "step": 11472
    },
    {
      "epoch": 0.8270020903914078,
      "grad_norm": 0.1821432237047736,
      "learning_rate": 1.5292109108797726e-05,
      "loss": 0.9903,
      "step": 11473
    },
    {
      "epoch": 0.8270741728537447,
      "grad_norm": 0.12510864838634309,
      "learning_rate": 1.5279703173925063e-05,
      "loss": 1.0743,
      "step": 11474
    },
    {
      "epoch": 0.8271462553160815,
      "grad_norm": 0.11967430434704575,
      "learning_rate": 1.5267301857066508e-05,
      "loss": 1.0137,
      "step": 11475
    },
    {
      "epoch": 0.8272183377784185,
      "grad_norm": 0.1427511368200765,
      "learning_rate": 1.5254905158898037e-05,
      "loss": 1.0547,
      "step": 11476
    },
    {
      "epoch": 0.8272904202407554,
      "grad_norm": 0.1231388677828343,
      "learning_rate": 1.5242513080095389e-05,
      "loss": 1.0146,
      "step": 11477
    },
    {
      "epoch": 0.8273625027030923,
      "grad_norm": 0.13155303282209574,
      "learning_rate": 1.5230125621334046e-05,
      "loss": 1.0954,
      "step": 11478
    },
    {
      "epoch": 0.8274345851654292,
      "grad_norm": 0.10774560179841233,
      "learning_rate": 1.5217742783289224e-05,
      "loss": 1.0747,
      "step": 11479
    },
    {
      "epoch": 0.8275066676277661,
      "grad_norm": 0.12257917460014048,
      "learning_rate": 1.5205364566635904e-05,
      "loss": 0.9994,
      "step": 11480
    },
    {
      "epoch": 0.8275787500901031,
      "grad_norm": 0.13008913890886198,
      "learning_rate": 1.5192990972048804e-05,
      "loss": 0.9994,
      "step": 11481
    },
    {
      "epoch": 0.82765083255244,
      "grad_norm": 0.10088433197762048,
      "learning_rate": 1.5180622000202416e-05,
      "loss": 1.0118,
      "step": 11482
    },
    {
      "epoch": 0.8277229150147769,
      "grad_norm": 0.28321294905652006,
      "learning_rate": 1.5168257651770934e-05,
      "loss": 0.9878,
      "step": 11483
    },
    {
      "epoch": 0.8277949974771138,
      "grad_norm": 0.11316066666319653,
      "learning_rate": 1.515589792742833e-05,
      "loss": 1.0273,
      "step": 11484
    },
    {
      "epoch": 0.8278670799394507,
      "grad_norm": 0.2750814930535907,
      "learning_rate": 1.5143542827848345e-05,
      "loss": 1.071,
      "step": 11485
    },
    {
      "epoch": 0.8279391624017877,
      "grad_norm": 0.12272954751455993,
      "learning_rate": 1.5131192353704404e-05,
      "loss": 0.9389,
      "step": 11486
    },
    {
      "epoch": 0.8280112448641246,
      "grad_norm": 0.12173163124626174,
      "learning_rate": 1.5118846505669748e-05,
      "loss": 1.0172,
      "step": 11487
    },
    {
      "epoch": 0.8280833273264615,
      "grad_norm": 0.12827346772349352,
      "learning_rate": 1.5106505284417327e-05,
      "loss": 1.0032,
      "step": 11488
    },
    {
      "epoch": 0.8281554097887984,
      "grad_norm": 0.1106624312463049,
      "learning_rate": 1.5094168690619848e-05,
      "loss": 1.0129,
      "step": 11489
    },
    {
      "epoch": 0.8282274922511353,
      "grad_norm": 0.19145126647852168,
      "learning_rate": 1.508183672494976e-05,
      "loss": 1.102,
      "step": 11490
    },
    {
      "epoch": 0.8282995747134723,
      "grad_norm": 0.11408844312817566,
      "learning_rate": 1.5069509388079273e-05,
      "loss": 1.0618,
      "step": 11491
    },
    {
      "epoch": 0.8283716571758091,
      "grad_norm": 0.13530613670678265,
      "learning_rate": 1.5057186680680336e-05,
      "loss": 1.0012,
      "step": 11492
    },
    {
      "epoch": 0.828443739638146,
      "grad_norm": 0.1162664071284391,
      "learning_rate": 1.5044868603424655e-05,
      "loss": 1.049,
      "step": 11493
    },
    {
      "epoch": 0.8285158221004829,
      "grad_norm": 0.11731302937882009,
      "learning_rate": 1.5032555156983629e-05,
      "loss": 1.0519,
      "step": 11494
    },
    {
      "epoch": 0.8285879045628198,
      "grad_norm": 1.5833158046498341,
      "learning_rate": 1.502024634202851e-05,
      "loss": 1.0186,
      "step": 11495
    },
    {
      "epoch": 0.8286599870251568,
      "grad_norm": 0.12286166024605416,
      "learning_rate": 1.5007942159230226e-05,
      "loss": 1.0269,
      "step": 11496
    },
    {
      "epoch": 0.8287320694874937,
      "grad_norm": 0.13664121526819742,
      "learning_rate": 1.4995642609259463e-05,
      "loss": 1.056,
      "step": 11497
    },
    {
      "epoch": 0.8288041519498306,
      "grad_norm": 0.11911840435209796,
      "learning_rate": 1.4983347692786609e-05,
      "loss": 1.0516,
      "step": 11498
    },
    {
      "epoch": 0.8288762344121675,
      "grad_norm": 0.11059210680205635,
      "learning_rate": 1.4971057410481904e-05,
      "loss": 1.0461,
      "step": 11499
    },
    {
      "epoch": 0.8289483168745044,
      "grad_norm": 0.10962014288304493,
      "learning_rate": 1.4958771763015267e-05,
      "loss": 1.0024,
      "step": 11500
    },
    {
      "epoch": 0.8290203993368414,
      "grad_norm": 0.11240724874659887,
      "learning_rate": 1.4946490751056397e-05,
      "loss": 0.999,
      "step": 11501
    },
    {
      "epoch": 0.8290924817991783,
      "grad_norm": 0.1378009228660086,
      "learning_rate": 1.4934214375274635e-05,
      "loss": 0.9941,
      "step": 11502
    },
    {
      "epoch": 0.8291645642615152,
      "grad_norm": 0.16622785000666926,
      "learning_rate": 1.4921942636339236e-05,
      "loss": 1.1462,
      "step": 11503
    },
    {
      "epoch": 0.8292366467238521,
      "grad_norm": 1.4163332333057819,
      "learning_rate": 1.4909675534919122e-05,
      "loss": 1.0469,
      "step": 11504
    },
    {
      "epoch": 0.829308729186189,
      "grad_norm": 0.16115508253422425,
      "learning_rate": 1.4897413071682897e-05,
      "loss": 1.0344,
      "step": 11505
    },
    {
      "epoch": 0.829380811648526,
      "grad_norm": 0.11487288564009203,
      "learning_rate": 1.488515524729902e-05,
      "loss": 1.0964,
      "step": 11506
    },
    {
      "epoch": 0.8294528941108629,
      "grad_norm": 0.2992924002778191,
      "learning_rate": 1.4872902062435612e-05,
      "loss": 1.0107,
      "step": 11507
    },
    {
      "epoch": 0.8295249765731998,
      "grad_norm": 0.12333539682072832,
      "learning_rate": 1.4860653517760659e-05,
      "loss": 1.0524,
      "step": 11508
    },
    {
      "epoch": 0.8295970590355366,
      "grad_norm": 0.10743078181361984,
      "learning_rate": 1.4848409613941738e-05,
      "loss": 0.9737,
      "step": 11509
    },
    {
      "epoch": 0.8296691414978735,
      "grad_norm": 0.16003947005001373,
      "learning_rate": 1.4836170351646294e-05,
      "loss": 0.9219,
      "step": 11510
    },
    {
      "epoch": 0.8297412239602104,
      "grad_norm": 0.12620426503959864,
      "learning_rate": 1.4823935731541427e-05,
      "loss": 1.0957,
      "step": 11511
    },
    {
      "epoch": 0.8298133064225474,
      "grad_norm": 0.12314018710987996,
      "learning_rate": 1.4811705754294125e-05,
      "loss": 1.0658,
      "step": 11512
    },
    {
      "epoch": 0.8298853888848843,
      "grad_norm": 0.12080329260188939,
      "learning_rate": 1.4799480420570943e-05,
      "loss": 1.0741,
      "step": 11513
    },
    {
      "epoch": 0.8299574713472212,
      "grad_norm": 0.12441203768338822,
      "learning_rate": 1.4787259731038317e-05,
      "loss": 0.9728,
      "step": 11514
    },
    {
      "epoch": 0.8300295538095581,
      "grad_norm": 0.12561603244365332,
      "learning_rate": 1.4775043686362367e-05,
      "loss": 0.995,
      "step": 11515
    },
    {
      "epoch": 0.830101636271895,
      "grad_norm": 0.1348325187714438,
      "learning_rate": 1.4762832287208972e-05,
      "loss": 1.0841,
      "step": 11516
    },
    {
      "epoch": 0.830173718734232,
      "grad_norm": 0.13248896126995485,
      "learning_rate": 1.4750625534243779e-05,
      "loss": 1.0289,
      "step": 11517
    },
    {
      "epoch": 0.8302458011965689,
      "grad_norm": 0.11444979425173503,
      "learning_rate": 1.4738423428132153e-05,
      "loss": 1.0655,
      "step": 11518
    },
    {
      "epoch": 0.8303178836589058,
      "grad_norm": 0.10894893454752214,
      "learning_rate": 1.472622596953922e-05,
      "loss": 1.0892,
      "step": 11519
    },
    {
      "epoch": 0.8303899661212427,
      "grad_norm": 0.1440698204036915,
      "learning_rate": 1.471403315912986e-05,
      "loss": 1.057,
      "step": 11520
    },
    {
      "epoch": 0.8304620485835796,
      "grad_norm": 0.1381035285148383,
      "learning_rate": 1.470184499756867e-05,
      "loss": 1.0332,
      "step": 11521
    },
    {
      "epoch": 0.8305341310459166,
      "grad_norm": 0.11295238206565252,
      "learning_rate": 1.4689661485520035e-05,
      "loss": 1.0781,
      "step": 11522
    },
    {
      "epoch": 0.8306062135082535,
      "grad_norm": 0.11733125842073241,
      "learning_rate": 1.4677482623648064e-05,
      "loss": 1.041,
      "step": 11523
    },
    {
      "epoch": 0.8306782959705904,
      "grad_norm": 0.12752988994034353,
      "learning_rate": 1.4665308412616596e-05,
      "loss": 1.0777,
      "step": 11524
    },
    {
      "epoch": 0.8307503784329273,
      "grad_norm": 0.2033766102417877,
      "learning_rate": 1.4653138853089254e-05,
      "loss": 1.0648,
      "step": 11525
    },
    {
      "epoch": 0.8308224608952642,
      "grad_norm": 0.12472757696479976,
      "learning_rate": 1.4640973945729374e-05,
      "loss": 1.112,
      "step": 11526
    },
    {
      "epoch": 0.830894543357601,
      "grad_norm": 0.12040959793938469,
      "learning_rate": 1.4628813691200049e-05,
      "loss": 1.0498,
      "step": 11527
    },
    {
      "epoch": 0.830966625819938,
      "grad_norm": 0.11247893478242696,
      "learning_rate": 1.4616658090164149e-05,
      "loss": 1.0321,
      "step": 11528
    },
    {
      "epoch": 0.8310387082822749,
      "grad_norm": 0.22913973805129131,
      "learning_rate": 1.460450714328423e-05,
      "loss": 1.0699,
      "step": 11529
    },
    {
      "epoch": 0.8311107907446118,
      "grad_norm": 0.1104955850889857,
      "learning_rate": 1.4592360851222653e-05,
      "loss": 1.027,
      "step": 11530
    },
    {
      "epoch": 0.8311828732069487,
      "grad_norm": 0.13057220383532706,
      "learning_rate": 1.4580219214641478e-05,
      "loss": 1.038,
      "step": 11531
    },
    {
      "epoch": 0.8312549556692856,
      "grad_norm": 0.1272602609173868,
      "learning_rate": 1.4568082234202551e-05,
      "loss": 1.1816,
      "step": 11532
    },
    {
      "epoch": 0.8313270381316226,
      "grad_norm": 0.11850830378790651,
      "learning_rate": 1.4555949910567435e-05,
      "loss": 0.9461,
      "step": 11533
    },
    {
      "epoch": 0.8313991205939595,
      "grad_norm": 0.19918718885469866,
      "learning_rate": 1.4543822244397443e-05,
      "loss": 1.0833,
      "step": 11534
    },
    {
      "epoch": 0.8314712030562964,
      "grad_norm": 0.11415277616737754,
      "learning_rate": 1.4531699236353669e-05,
      "loss": 1.0933,
      "step": 11535
    },
    {
      "epoch": 0.8315432855186333,
      "grad_norm": 0.11483393074330113,
      "learning_rate": 1.4519580887096895e-05,
      "loss": 0.9578,
      "step": 11536
    },
    {
      "epoch": 0.8316153679809702,
      "grad_norm": 0.12922295965902317,
      "learning_rate": 1.4507467197287717e-05,
      "loss": 1.0587,
      "step": 11537
    },
    {
      "epoch": 0.8316874504433072,
      "grad_norm": 0.15646848465371233,
      "learning_rate": 1.449535816758637e-05,
      "loss": 0.9879,
      "step": 11538
    },
    {
      "epoch": 0.8317595329056441,
      "grad_norm": 0.11753302456989574,
      "learning_rate": 1.4483253798652974e-05,
      "loss": 0.9887,
      "step": 11539
    },
    {
      "epoch": 0.831831615367981,
      "grad_norm": 0.14514727185968765,
      "learning_rate": 1.447115409114731e-05,
      "loss": 1.0575,
      "step": 11540
    },
    {
      "epoch": 0.8319036978303179,
      "grad_norm": 0.12452011448815854,
      "learning_rate": 1.4459059045728918e-05,
      "loss": 1.05,
      "step": 11541
    },
    {
      "epoch": 0.8319757802926548,
      "grad_norm": 0.11188883090542805,
      "learning_rate": 1.4446968663057037e-05,
      "loss": 1.0405,
      "step": 11542
    },
    {
      "epoch": 0.8320478627549918,
      "grad_norm": 0.1183750017504905,
      "learning_rate": 1.4434882943790773e-05,
      "loss": 1.0716,
      "step": 11543
    },
    {
      "epoch": 0.8321199452173286,
      "grad_norm": 0.12695267518841138,
      "learning_rate": 1.4422801888588866e-05,
      "loss": 0.9879,
      "step": 11544
    },
    {
      "epoch": 0.8321920276796655,
      "grad_norm": 0.30873307312901727,
      "learning_rate": 1.4410725498109889e-05,
      "loss": 1.0475,
      "step": 11545
    },
    {
      "epoch": 0.8322641101420024,
      "grad_norm": 0.11689178894897845,
      "learning_rate": 1.4398653773012017e-05,
      "loss": 1.0614,
      "step": 11546
    },
    {
      "epoch": 0.8323361926043393,
      "grad_norm": 0.15998360051460703,
      "learning_rate": 1.4386586713953353e-05,
      "loss": 1.0347,
      "step": 11547
    },
    {
      "epoch": 0.8324082750666763,
      "grad_norm": 0.12209387779028946,
      "learning_rate": 1.4374524321591632e-05,
      "loss": 1.0057,
      "step": 11548
    },
    {
      "epoch": 0.8324803575290132,
      "grad_norm": 0.12763595633743505,
      "learning_rate": 1.4362466596584389e-05,
      "loss": 1.048,
      "step": 11549
    },
    {
      "epoch": 0.8325524399913501,
      "grad_norm": 0.11547999636177123,
      "learning_rate": 1.4350413539588825e-05,
      "loss": 1.0364,
      "step": 11550
    },
    {
      "epoch": 0.832624522453687,
      "grad_norm": 0.11443611473625807,
      "learning_rate": 1.4338365151261945e-05,
      "loss": 1.0464,
      "step": 11551
    },
    {
      "epoch": 0.8326966049160239,
      "grad_norm": 0.13053614799225413,
      "learning_rate": 1.4326321432260537e-05,
      "loss": 1.0265,
      "step": 11552
    },
    {
      "epoch": 0.8327686873783609,
      "grad_norm": 0.12048745919185229,
      "learning_rate": 1.4314282383241096e-05,
      "loss": 1.0879,
      "step": 11553
    },
    {
      "epoch": 0.8328407698406978,
      "grad_norm": 0.11189244136227107,
      "learning_rate": 1.4302248004859809e-05,
      "loss": 1.0742,
      "step": 11554
    },
    {
      "epoch": 0.8329128523030347,
      "grad_norm": 0.11495667889173439,
      "learning_rate": 1.429021829777265e-05,
      "loss": 1.0286,
      "step": 11555
    },
    {
      "epoch": 0.8329849347653716,
      "grad_norm": 0.12522831725127206,
      "learning_rate": 1.4278193262635397e-05,
      "loss": 1.062,
      "step": 11556
    },
    {
      "epoch": 0.8330570172277085,
      "grad_norm": 0.12572520972992954,
      "learning_rate": 1.4266172900103525e-05,
      "loss": 1.0639,
      "step": 11557
    },
    {
      "epoch": 0.8331290996900454,
      "grad_norm": 0.11827221095249933,
      "learning_rate": 1.4254157210832208e-05,
      "loss": 0.9755,
      "step": 11558
    },
    {
      "epoch": 0.8332011821523824,
      "grad_norm": 0.11142935894674144,
      "learning_rate": 1.4242146195476425e-05,
      "loss": 1.003,
      "step": 11559
    },
    {
      "epoch": 0.8332732646147193,
      "grad_norm": 0.157706336091132,
      "learning_rate": 1.4230139854690872e-05,
      "loss": 1.0281,
      "step": 11560
    },
    {
      "epoch": 0.8333453470770561,
      "grad_norm": 0.14641734484305247,
      "learning_rate": 1.4218138189130047e-05,
      "loss": 1.0683,
      "step": 11561
    },
    {
      "epoch": 0.833417429539393,
      "grad_norm": 0.11866756039074287,
      "learning_rate": 1.4206141199448109e-05,
      "loss": 1.0219,
      "step": 11562
    },
    {
      "epoch": 0.8334895120017299,
      "grad_norm": 0.11663809604910066,
      "learning_rate": 1.4194148886298997e-05,
      "loss": 1.0829,
      "step": 11563
    },
    {
      "epoch": 0.8335615944640669,
      "grad_norm": 0.15849623818187783,
      "learning_rate": 1.4182161250336389e-05,
      "loss": 0.9632,
      "step": 11564
    },
    {
      "epoch": 0.8336336769264038,
      "grad_norm": 0.18378845456646237,
      "learning_rate": 1.4170178292213799e-05,
      "loss": 1.0523,
      "step": 11565
    },
    {
      "epoch": 0.8337057593887407,
      "grad_norm": 0.11807028684141274,
      "learning_rate": 1.415820001258431e-05,
      "loss": 1.018,
      "step": 11566
    },
    {
      "epoch": 0.8337778418510776,
      "grad_norm": 0.1141796189185032,
      "learning_rate": 1.4146226412100893e-05,
      "loss": 1.0829,
      "step": 11567
    },
    {
      "epoch": 0.8338499243134145,
      "grad_norm": 0.14340766268056296,
      "learning_rate": 1.413425749141618e-05,
      "loss": 1.0431,
      "step": 11568
    },
    {
      "epoch": 0.8339220067757515,
      "grad_norm": 0.1104940639122175,
      "learning_rate": 1.4122293251182661e-05,
      "loss": 1.0813,
      "step": 11569
    },
    {
      "epoch": 0.8339940892380884,
      "grad_norm": 0.1223994414114709,
      "learning_rate": 1.4110333692052413e-05,
      "loss": 1.14,
      "step": 11570
    },
    {
      "epoch": 0.8340661717004253,
      "grad_norm": 0.1369429813023566,
      "learning_rate": 1.4098378814677382e-05,
      "loss": 1.0048,
      "step": 11571
    },
    {
      "epoch": 0.8341382541627622,
      "grad_norm": 0.16138593051518188,
      "learning_rate": 1.408642861970919e-05,
      "loss": 1.0471,
      "step": 11572
    },
    {
      "epoch": 0.8342103366250991,
      "grad_norm": 0.13222275487672022,
      "learning_rate": 1.4074483107799253e-05,
      "loss": 1.0203,
      "step": 11573
    },
    {
      "epoch": 0.8342824190874361,
      "grad_norm": 0.11356617736443976,
      "learning_rate": 1.4062542279598689e-05,
      "loss": 0.9418,
      "step": 11574
    },
    {
      "epoch": 0.834354501549773,
      "grad_norm": 0.11393180553827416,
      "learning_rate": 1.4050606135758392e-05,
      "loss": 1.0026,
      "step": 11575
    },
    {
      "epoch": 0.8344265840121099,
      "grad_norm": 0.12865229713724197,
      "learning_rate": 1.4038674676928986e-05,
      "loss": 1.1273,
      "step": 11576
    },
    {
      "epoch": 0.8344986664744468,
      "grad_norm": 0.12284307279826352,
      "learning_rate": 1.402674790376085e-05,
      "loss": 1.0333,
      "step": 11577
    },
    {
      "epoch": 0.8345707489367837,
      "grad_norm": 0.14100487754673824,
      "learning_rate": 1.4014825816904075e-05,
      "loss": 1.1025,
      "step": 11578
    },
    {
      "epoch": 0.8346428313991205,
      "grad_norm": 0.25089941863198045,
      "learning_rate": 1.4002908417008553e-05,
      "loss": 1.0294,
      "step": 11579
    },
    {
      "epoch": 0.8347149138614575,
      "grad_norm": 0.12728191276320788,
      "learning_rate": 1.3990995704723875e-05,
      "loss": 1.0657,
      "step": 11580
    },
    {
      "epoch": 0.8347869963237944,
      "grad_norm": 0.1184919973715827,
      "learning_rate": 1.3979087680699377e-05,
      "loss": 1.0531,
      "step": 11581
    },
    {
      "epoch": 0.8348590787861313,
      "grad_norm": 0.17845302464597113,
      "learning_rate": 1.3967184345584173e-05,
      "loss": 1.0137,
      "step": 11582
    },
    {
      "epoch": 0.8349311612484682,
      "grad_norm": 0.15207349742790474,
      "learning_rate": 1.3955285700027099e-05,
      "loss": 1.1264,
      "step": 11583
    },
    {
      "epoch": 0.8350032437108051,
      "grad_norm": 0.12124384434103991,
      "learning_rate": 1.3943391744676727e-05,
      "loss": 1.0493,
      "step": 11584
    },
    {
      "epoch": 0.8350753261731421,
      "grad_norm": 0.13911912747877056,
      "learning_rate": 1.3931502480181391e-05,
      "loss": 1.0065,
      "step": 11585
    },
    {
      "epoch": 0.835147408635479,
      "grad_norm": 0.12313709499231469,
      "learning_rate": 1.3919617907189154e-05,
      "loss": 1.0476,
      "step": 11586
    },
    {
      "epoch": 0.8352194910978159,
      "grad_norm": 0.11775608848911963,
      "learning_rate": 1.3907738026347849e-05,
      "loss": 0.9833,
      "step": 11587
    },
    {
      "epoch": 0.8352915735601528,
      "grad_norm": 0.14760574675820937,
      "learning_rate": 1.3895862838305018e-05,
      "loss": 1.0511,
      "step": 11588
    },
    {
      "epoch": 0.8353636560224897,
      "grad_norm": 0.1271224342009122,
      "learning_rate": 1.3883992343707975e-05,
      "loss": 1.0402,
      "step": 11589
    },
    {
      "epoch": 0.8354357384848267,
      "grad_norm": 0.11923145487775114,
      "learning_rate": 1.3872126543203755e-05,
      "loss": 1.0085,
      "step": 11590
    },
    {
      "epoch": 0.8355078209471636,
      "grad_norm": 0.10803316470200863,
      "learning_rate": 1.3860265437439169e-05,
      "loss": 1.0321,
      "step": 11591
    },
    {
      "epoch": 0.8355799034095005,
      "grad_norm": 0.13479713757596243,
      "learning_rate": 1.3848409027060748e-05,
      "loss": 1.0927,
      "step": 11592
    },
    {
      "epoch": 0.8356519858718374,
      "grad_norm": 0.12249168725059403,
      "learning_rate": 1.383655731271477e-05,
      "loss": 1.0452,
      "step": 11593
    },
    {
      "epoch": 0.8357240683341743,
      "grad_norm": 0.1251410943665959,
      "learning_rate": 1.3824710295047272e-05,
      "loss": 1.0024,
      "step": 11594
    },
    {
      "epoch": 0.8357961507965113,
      "grad_norm": 0.1310479869057059,
      "learning_rate": 1.381286797470397e-05,
      "loss": 1.0743,
      "step": 11595
    },
    {
      "epoch": 0.8358682332588481,
      "grad_norm": 0.1565247415716367,
      "learning_rate": 1.3801030352330435e-05,
      "loss": 1.0886,
      "step": 11596
    },
    {
      "epoch": 0.835940315721185,
      "grad_norm": 0.2556520627166862,
      "learning_rate": 1.3789197428571909e-05,
      "loss": 0.9929,
      "step": 11597
    },
    {
      "epoch": 0.8360123981835219,
      "grad_norm": 0.12076566264119168,
      "learning_rate": 1.377736920407341e-05,
      "loss": 1.0379,
      "step": 11598
    },
    {
      "epoch": 0.8360844806458588,
      "grad_norm": 0.12082569514138262,
      "learning_rate": 1.3765545679479609e-05,
      "loss": 1.0086,
      "step": 11599
    },
    {
      "epoch": 0.8361565631081957,
      "grad_norm": 0.12337399505333342,
      "learning_rate": 1.3753726855435068e-05,
      "loss": 0.9616,
      "step": 11600
    },
    {
      "epoch": 0.8362286455705327,
      "grad_norm": 0.11147313389589772,
      "learning_rate": 1.3741912732583994e-05,
      "loss": 1.0483,
      "step": 11601
    },
    {
      "epoch": 0.8363007280328696,
      "grad_norm": 0.14873622803346967,
      "learning_rate": 1.3730103311570386e-05,
      "loss": 1.0473,
      "step": 11602
    },
    {
      "epoch": 0.8363728104952065,
      "grad_norm": 0.13887355779952068,
      "learning_rate": 1.3718298593037916e-05,
      "loss": 1.048,
      "step": 11603
    },
    {
      "epoch": 0.8364448929575434,
      "grad_norm": 0.1482210927415912,
      "learning_rate": 1.3706498577630045e-05,
      "loss": 1.0022,
      "step": 11604
    },
    {
      "epoch": 0.8365169754198803,
      "grad_norm": 0.11753717352760196,
      "learning_rate": 1.3694703265990039e-05,
      "loss": 1.0174,
      "step": 11605
    },
    {
      "epoch": 0.8365890578822173,
      "grad_norm": 0.1257116329058317,
      "learning_rate": 1.3682912658760827e-05,
      "loss": 1.0042,
      "step": 11606
    },
    {
      "epoch": 0.8366611403445542,
      "grad_norm": 0.12286363487373896,
      "learning_rate": 1.3671126756585073e-05,
      "loss": 1.0055,
      "step": 11607
    },
    {
      "epoch": 0.8367332228068911,
      "grad_norm": 0.1183384593046308,
      "learning_rate": 1.3659345560105196e-05,
      "loss": 1.0585,
      "step": 11608
    },
    {
      "epoch": 0.836805305269228,
      "grad_norm": 0.11649614975786235,
      "learning_rate": 1.364756906996345e-05,
      "loss": 0.9809,
      "step": 11609
    },
    {
      "epoch": 0.836877387731565,
      "grad_norm": 0.11227413575093777,
      "learning_rate": 1.3635797286801733e-05,
      "loss": 1.0503,
      "step": 11610
    },
    {
      "epoch": 0.8369494701939019,
      "grad_norm": 0.12211056866234385,
      "learning_rate": 1.3624030211261685e-05,
      "loss": 1.0027,
      "step": 11611
    },
    {
      "epoch": 0.8370215526562388,
      "grad_norm": 0.12020633152816217,
      "learning_rate": 1.3612267843984706e-05,
      "loss": 1.0536,
      "step": 11612
    },
    {
      "epoch": 0.8370936351185756,
      "grad_norm": 0.11584832572968654,
      "learning_rate": 1.3600510185612004e-05,
      "loss": 0.9947,
      "step": 11613
    },
    {
      "epoch": 0.8371657175809125,
      "grad_norm": 0.11798707835749686,
      "learning_rate": 1.3588757236784456e-05,
      "loss": 1.0406,
      "step": 11614
    },
    {
      "epoch": 0.8372378000432494,
      "grad_norm": 0.1159603369177336,
      "learning_rate": 1.3577008998142692e-05,
      "loss": 1.0601,
      "step": 11615
    },
    {
      "epoch": 0.8373098825055864,
      "grad_norm": 0.1159310948802757,
      "learning_rate": 1.3565265470327082e-05,
      "loss": 1.0422,
      "step": 11616
    },
    {
      "epoch": 0.8373819649679233,
      "grad_norm": 0.11174277924810513,
      "learning_rate": 1.3553526653977766e-05,
      "loss": 1.0145,
      "step": 11617
    },
    {
      "epoch": 0.8374540474302602,
      "grad_norm": 0.13077308171367685,
      "learning_rate": 1.3541792549734655e-05,
      "loss": 1.0079,
      "step": 11618
    },
    {
      "epoch": 0.8375261298925971,
      "grad_norm": 0.10343330953232618,
      "learning_rate": 1.3530063158237305e-05,
      "loss": 1.0587,
      "step": 11619
    },
    {
      "epoch": 0.837598212354934,
      "grad_norm": 0.16305623512223327,
      "learning_rate": 1.3518338480125093e-05,
      "loss": 1.0293,
      "step": 11620
    },
    {
      "epoch": 0.837670294817271,
      "grad_norm": 0.1326079692143046,
      "learning_rate": 1.3506618516037106e-05,
      "loss": 1.1751,
      "step": 11621
    },
    {
      "epoch": 0.8377423772796079,
      "grad_norm": 0.1678310079354567,
      "learning_rate": 1.3494903266612236e-05,
      "loss": 0.9791,
      "step": 11622
    },
    {
      "epoch": 0.8378144597419448,
      "grad_norm": 0.11070434312433611,
      "learning_rate": 1.3483192732489025e-05,
      "loss": 1.0266,
      "step": 11623
    },
    {
      "epoch": 0.8378865422042817,
      "grad_norm": 0.11729378499303646,
      "learning_rate": 1.3471486914305798e-05,
      "loss": 1.0419,
      "step": 11624
    },
    {
      "epoch": 0.8379586246666186,
      "grad_norm": 0.1146086795233956,
      "learning_rate": 1.3459785812700654e-05,
      "loss": 1.0243,
      "step": 11625
    },
    {
      "epoch": 0.8380307071289556,
      "grad_norm": 0.11621935123849227,
      "learning_rate": 1.3448089428311384e-05,
      "loss": 1.0562,
      "step": 11626
    },
    {
      "epoch": 0.8381027895912925,
      "grad_norm": 0.22728902916932653,
      "learning_rate": 1.3436397761775565e-05,
      "loss": 1.0535,
      "step": 11627
    },
    {
      "epoch": 0.8381748720536294,
      "grad_norm": 0.12388194745078501,
      "learning_rate": 1.342471081373048e-05,
      "loss": 1.009,
      "step": 11628
    },
    {
      "epoch": 0.8382469545159663,
      "grad_norm": 0.4514396745505334,
      "learning_rate": 1.341302858481318e-05,
      "loss": 1.1275,
      "step": 11629
    },
    {
      "epoch": 0.8383190369783032,
      "grad_norm": 0.8717686272321852,
      "learning_rate": 1.3401351075660452e-05,
      "loss": 1.0375,
      "step": 11630
    },
    {
      "epoch": 0.83839111944064,
      "grad_norm": 0.15638962190742325,
      "learning_rate": 1.3389678286908824e-05,
      "loss": 1.0286,
      "step": 11631
    },
    {
      "epoch": 0.838463201902977,
      "grad_norm": 0.13869530739601915,
      "learning_rate": 1.337801021919457e-05,
      "loss": 1.0677,
      "step": 11632
    },
    {
      "epoch": 0.8385352843653139,
      "grad_norm": 0.12359701478163297,
      "learning_rate": 1.33663468731537e-05,
      "loss": 1.0377,
      "step": 11633
    },
    {
      "epoch": 0.8386073668276508,
      "grad_norm": 0.13483475229345718,
      "learning_rate": 1.3354688249421977e-05,
      "loss": 1.004,
      "step": 11634
    },
    {
      "epoch": 0.8386794492899877,
      "grad_norm": 0.14047156178390086,
      "learning_rate": 1.3343034348634886e-05,
      "loss": 1.0833,
      "step": 11635
    },
    {
      "epoch": 0.8387515317523246,
      "grad_norm": 0.12456383507746835,
      "learning_rate": 1.3331385171427691e-05,
      "loss": 1.0193,
      "step": 11636
    },
    {
      "epoch": 0.8388236142146616,
      "grad_norm": 0.11478334464699295,
      "learning_rate": 1.3319740718435358e-05,
      "loss": 1.0831,
      "step": 11637
    },
    {
      "epoch": 0.8388956966769985,
      "grad_norm": 0.10942992031400527,
      "learning_rate": 1.3308100990292616e-05,
      "loss": 1.0264,
      "step": 11638
    },
    {
      "epoch": 0.8389677791393354,
      "grad_norm": 0.42362133880267494,
      "learning_rate": 1.3296465987633943e-05,
      "loss": 0.9604,
      "step": 11639
    },
    {
      "epoch": 0.8390398616016723,
      "grad_norm": 0.12049477534537824,
      "learning_rate": 1.3284835711093535e-05,
      "loss": 1.0401,
      "step": 11640
    },
    {
      "epoch": 0.8391119440640092,
      "grad_norm": 0.133813088850917,
      "learning_rate": 1.3273210161305372e-05,
      "loss": 1.0602,
      "step": 11641
    },
    {
      "epoch": 0.8391840265263462,
      "grad_norm": 0.11701050236075021,
      "learning_rate": 1.3261589338903124e-05,
      "loss": 1.0391,
      "step": 11642
    },
    {
      "epoch": 0.8392561089886831,
      "grad_norm": 0.1098800514494914,
      "learning_rate": 1.3249973244520241e-05,
      "loss": 1.0719,
      "step": 11643
    },
    {
      "epoch": 0.83932819145102,
      "grad_norm": 0.11311976811776137,
      "learning_rate": 1.3238361878789919e-05,
      "loss": 1.0547,
      "step": 11644
    },
    {
      "epoch": 0.8394002739133569,
      "grad_norm": 0.15435223359427439,
      "learning_rate": 1.3226755242345046e-05,
      "loss": 1.0549,
      "step": 11645
    },
    {
      "epoch": 0.8394723563756938,
      "grad_norm": 0.12789204399043735,
      "learning_rate": 1.321515333581832e-05,
      "loss": 1.0147,
      "step": 11646
    },
    {
      "epoch": 0.8395444388380308,
      "grad_norm": 0.14128065915051774,
      "learning_rate": 1.3203556159842156e-05,
      "loss": 1.0519,
      "step": 11647
    },
    {
      "epoch": 0.8396165213003676,
      "grad_norm": 0.11470123675863292,
      "learning_rate": 1.3191963715048638e-05,
      "loss": 1.0784,
      "step": 11648
    },
    {
      "epoch": 0.8396886037627045,
      "grad_norm": 0.13464027023804379,
      "learning_rate": 1.3180376002069728e-05,
      "loss": 1.0235,
      "step": 11649
    },
    {
      "epoch": 0.8397606862250414,
      "grad_norm": 0.12197211041827068,
      "learning_rate": 1.316879302153704e-05,
      "loss": 1.0446,
      "step": 11650
    },
    {
      "epoch": 0.8398327686873783,
      "grad_norm": 0.10960255311812908,
      "learning_rate": 1.3157214774081972e-05,
      "loss": 1.0208,
      "step": 11651
    },
    {
      "epoch": 0.8399048511497152,
      "grad_norm": 0.11791626009168613,
      "learning_rate": 1.3145641260335562e-05,
      "loss": 1.028,
      "step": 11652
    },
    {
      "epoch": 0.8399769336120522,
      "grad_norm": 0.12208248054860109,
      "learning_rate": 1.313407248092875e-05,
      "loss": 0.9707,
      "step": 11653
    },
    {
      "epoch": 0.8400490160743891,
      "grad_norm": 0.11323903469315473,
      "learning_rate": 1.3122508436492109e-05,
      "loss": 1.083,
      "step": 11654
    },
    {
      "epoch": 0.840121098536726,
      "grad_norm": 0.13778203211182016,
      "learning_rate": 1.3110949127656014e-05,
      "loss": 1.0476,
      "step": 11655
    },
    {
      "epoch": 0.8401931809990629,
      "grad_norm": 0.11604642522366228,
      "learning_rate": 1.3099394555050504e-05,
      "loss": 1.0206,
      "step": 11656
    },
    {
      "epoch": 0.8402652634613998,
      "grad_norm": 0.12537223414107015,
      "learning_rate": 1.3087844719305397e-05,
      "loss": 1.0178,
      "step": 11657
    },
    {
      "epoch": 0.8403373459237368,
      "grad_norm": 0.11383199782424813,
      "learning_rate": 1.3076299621050336e-05,
      "loss": 1.0096,
      "step": 11658
    },
    {
      "epoch": 0.8404094283860737,
      "grad_norm": 0.1421399577698599,
      "learning_rate": 1.3064759260914573e-05,
      "loss": 0.9992,
      "step": 11659
    },
    {
      "epoch": 0.8404815108484106,
      "grad_norm": 0.12612439308072135,
      "learning_rate": 1.3053223639527168e-05,
      "loss": 1.0923,
      "step": 11660
    },
    {
      "epoch": 0.8405535933107475,
      "grad_norm": 0.25605765080042464,
      "learning_rate": 1.3041692757516898e-05,
      "loss": 1.1041,
      "step": 11661
    },
    {
      "epoch": 0.8406256757730844,
      "grad_norm": 0.11289262491349487,
      "learning_rate": 1.3030166615512374e-05,
      "loss": 0.9893,
      "step": 11662
    },
    {
      "epoch": 0.8406977582354214,
      "grad_norm": 0.12032900143488635,
      "learning_rate": 1.30186452141418e-05,
      "loss": 1.0656,
      "step": 11663
    },
    {
      "epoch": 0.8407698406977583,
      "grad_norm": 0.114225844340313,
      "learning_rate": 1.3007128554033222e-05,
      "loss": 0.9748,
      "step": 11664
    },
    {
      "epoch": 0.8408419231600951,
      "grad_norm": 0.11261591546631827,
      "learning_rate": 1.2995616635814379e-05,
      "loss": 1.0019,
      "step": 11665
    },
    {
      "epoch": 0.840914005622432,
      "grad_norm": 0.12531333992844876,
      "learning_rate": 1.2984109460112825e-05,
      "loss": 1.0429,
      "step": 11666
    },
    {
      "epoch": 0.8409860880847689,
      "grad_norm": 0.13007503854182897,
      "learning_rate": 1.2972607027555761e-05,
      "loss": 1.0351,
      "step": 11667
    },
    {
      "epoch": 0.8410581705471059,
      "grad_norm": 0.13016768413963098,
      "learning_rate": 1.2961109338770172e-05,
      "loss": 1.0776,
      "step": 11668
    },
    {
      "epoch": 0.8411302530094428,
      "grad_norm": 0.11921745147379405,
      "learning_rate": 1.2949616394382802e-05,
      "loss": 1.0439,
      "step": 11669
    },
    {
      "epoch": 0.8412023354717797,
      "grad_norm": 0.14920751530832738,
      "learning_rate": 1.2938128195020116e-05,
      "loss": 1.0662,
      "step": 11670
    },
    {
      "epoch": 0.8412744179341166,
      "grad_norm": 0.10528494425576855,
      "learning_rate": 1.2926644741308326e-05,
      "loss": 0.9452,
      "step": 11671
    },
    {
      "epoch": 0.8413465003964535,
      "grad_norm": 0.1466108806590454,
      "learning_rate": 1.2915166033873371e-05,
      "loss": 1.0384,
      "step": 11672
    },
    {
      "epoch": 0.8414185828587905,
      "grad_norm": 0.11673306457154549,
      "learning_rate": 1.2903692073340945e-05,
      "loss": 0.9913,
      "step": 11673
    },
    {
      "epoch": 0.8414906653211274,
      "grad_norm": 0.1263091266552603,
      "learning_rate": 1.289222286033649e-05,
      "loss": 1.0177,
      "step": 11674
    },
    {
      "epoch": 0.8415627477834643,
      "grad_norm": 0.1193579522282906,
      "learning_rate": 1.2880758395485183e-05,
      "loss": 1.0376,
      "step": 11675
    },
    {
      "epoch": 0.8416348302458012,
      "grad_norm": 0.11636556205454039,
      "learning_rate": 1.2869298679411934e-05,
      "loss": 0.9758,
      "step": 11676
    },
    {
      "epoch": 0.8417069127081381,
      "grad_norm": 0.12464643812704189,
      "learning_rate": 1.2857843712741401e-05,
      "loss": 1.064,
      "step": 11677
    },
    {
      "epoch": 0.841778995170475,
      "grad_norm": 0.12468072516174103,
      "learning_rate": 1.2846393496097975e-05,
      "loss": 1.0051,
      "step": 11678
    },
    {
      "epoch": 0.841851077632812,
      "grad_norm": 0.10996821917993206,
      "learning_rate": 1.2834948030105808e-05,
      "loss": 0.9774,
      "step": 11679
    },
    {
      "epoch": 0.8419231600951489,
      "grad_norm": 0.1344722705609677,
      "learning_rate": 1.2823507315388772e-05,
      "loss": 1.0228,
      "step": 11680
    },
    {
      "epoch": 0.8419952425574858,
      "grad_norm": 0.12364142234149285,
      "learning_rate": 1.2812071352570488e-05,
      "loss": 1.0243,
      "step": 11681
    },
    {
      "epoch": 0.8420673250198227,
      "grad_norm": 0.13110038468963334,
      "learning_rate": 1.2800640142274324e-05,
      "loss": 0.9953,
      "step": 11682
    },
    {
      "epoch": 0.8421394074821595,
      "grad_norm": 0.15278177271405158,
      "learning_rate": 1.278921368512338e-05,
      "loss": 1.0085,
      "step": 11683
    },
    {
      "epoch": 0.8422114899444965,
      "grad_norm": 0.13583803360743063,
      "learning_rate": 1.2777791981740495e-05,
      "loss": 1.0279,
      "step": 11684
    },
    {
      "epoch": 0.8422835724068334,
      "grad_norm": 0.2641636265831282,
      "learning_rate": 1.2766375032748268e-05,
      "loss": 1.082,
      "step": 11685
    },
    {
      "epoch": 0.8423556548691703,
      "grad_norm": 0.14507994815869438,
      "learning_rate": 1.2754962838769013e-05,
      "loss": 1.0443,
      "step": 11686
    },
    {
      "epoch": 0.8424277373315072,
      "grad_norm": 0.1158818748486,
      "learning_rate": 1.27435554004248e-05,
      "loss": 1.0021,
      "step": 11687
    },
    {
      "epoch": 0.8424998197938441,
      "grad_norm": 0.13340791890562625,
      "learning_rate": 1.2732152718337431e-05,
      "loss": 0.9634,
      "step": 11688
    },
    {
      "epoch": 0.8425719022561811,
      "grad_norm": 0.1524081470124838,
      "learning_rate": 1.2720754793128464e-05,
      "loss": 1.0293,
      "step": 11689
    },
    {
      "epoch": 0.842643984718518,
      "grad_norm": 0.11210375869745806,
      "learning_rate": 1.2709361625419181e-05,
      "loss": 1.0185,
      "step": 11690
    },
    {
      "epoch": 0.8427160671808549,
      "grad_norm": 0.12085641692658043,
      "learning_rate": 1.2697973215830638e-05,
      "loss": 1.054,
      "step": 11691
    },
    {
      "epoch": 0.8427881496431918,
      "grad_norm": 0.11222792819886332,
      "learning_rate": 1.268658956498353e-05,
      "loss": 1.0182,
      "step": 11692
    },
    {
      "epoch": 0.8428602321055287,
      "grad_norm": 0.11482256792535193,
      "learning_rate": 1.267521067349845e-05,
      "loss": 1.0143,
      "step": 11693
    },
    {
      "epoch": 0.8429323145678657,
      "grad_norm": 0.12486463945431195,
      "learning_rate": 1.2663836541995611e-05,
      "loss": 1.0378,
      "step": 11694
    },
    {
      "epoch": 0.8430043970302026,
      "grad_norm": 0.1159395430819501,
      "learning_rate": 1.2652467171095039e-05,
      "loss": 1.0298,
      "step": 11695
    },
    {
      "epoch": 0.8430764794925395,
      "grad_norm": 0.12276284958106042,
      "learning_rate": 1.2641102561416396e-05,
      "loss": 1.0077,
      "step": 11696
    },
    {
      "epoch": 0.8431485619548764,
      "grad_norm": 0.13273781833755388,
      "learning_rate": 1.2629742713579219e-05,
      "loss": 1.0371,
      "step": 11697
    },
    {
      "epoch": 0.8432206444172133,
      "grad_norm": 0.11979458574297278,
      "learning_rate": 1.2618387628202699e-05,
      "loss": 1.0178,
      "step": 11698
    },
    {
      "epoch": 0.8432927268795503,
      "grad_norm": 0.1232762929007818,
      "learning_rate": 1.2607037305905811e-05,
      "loss": 1.0396,
      "step": 11699
    },
    {
      "epoch": 0.8433648093418871,
      "grad_norm": 0.11825599640138444,
      "learning_rate": 1.2595691747307214e-05,
      "loss": 1.0337,
      "step": 11700
    },
    {
      "epoch": 0.843436891804224,
      "grad_norm": 0.10724022018121682,
      "learning_rate": 1.2584350953025337e-05,
      "loss": 1.0626,
      "step": 11701
    },
    {
      "epoch": 0.8435089742665609,
      "grad_norm": 0.12038027199217774,
      "learning_rate": 1.2573014923678395e-05,
      "loss": 0.9997,
      "step": 11702
    },
    {
      "epoch": 0.8435810567288978,
      "grad_norm": 0.11137587501683568,
      "learning_rate": 1.2561683659884305e-05,
      "loss": 0.9763,
      "step": 11703
    },
    {
      "epoch": 0.8436531391912347,
      "grad_norm": 0.12747328198153549,
      "learning_rate": 1.2550357162260672e-05,
      "loss": 0.9865,
      "step": 11704
    },
    {
      "epoch": 0.8437252216535717,
      "grad_norm": 0.1261911813788713,
      "learning_rate": 1.2539035431424917e-05,
      "loss": 1.0445,
      "step": 11705
    },
    {
      "epoch": 0.8437973041159086,
      "grad_norm": 0.12541553753313114,
      "learning_rate": 1.2527718467994187e-05,
      "loss": 1.0647,
      "step": 11706
    },
    {
      "epoch": 0.8438693865782455,
      "grad_norm": 0.1136066694700602,
      "learning_rate": 1.2516406272585368e-05,
      "loss": 1.1028,
      "step": 11707
    },
    {
      "epoch": 0.8439414690405824,
      "grad_norm": 0.34738815207301027,
      "learning_rate": 1.2505098845815044e-05,
      "loss": 1.0982,
      "step": 11708
    },
    {
      "epoch": 0.8440135515029193,
      "grad_norm": 0.12325860918312015,
      "learning_rate": 1.2493796188299567e-05,
      "loss": 1.0421,
      "step": 11709
    },
    {
      "epoch": 0.8440856339652563,
      "grad_norm": 0.12474942019722346,
      "learning_rate": 1.2482498300655065e-05,
      "loss": 1.0381,
      "step": 11710
    },
    {
      "epoch": 0.8441577164275932,
      "grad_norm": 0.1607544768593463,
      "learning_rate": 1.2471205183497381e-05,
      "loss": 1.0795,
      "step": 11711
    },
    {
      "epoch": 0.8442297988899301,
      "grad_norm": 0.11610302149653746,
      "learning_rate": 1.2459916837442054e-05,
      "loss": 1.023,
      "step": 11712
    },
    {
      "epoch": 0.844301881352267,
      "grad_norm": 0.13428609238437814,
      "learning_rate": 1.2448633263104415e-05,
      "loss": 1.1047,
      "step": 11713
    },
    {
      "epoch": 0.844373963814604,
      "grad_norm": 0.14787786719882665,
      "learning_rate": 1.2437354461099493e-05,
      "loss": 1.0225,
      "step": 11714
    },
    {
      "epoch": 0.8444460462769409,
      "grad_norm": 0.10353652302978263,
      "learning_rate": 1.2426080432042153e-05,
      "loss": 0.9827,
      "step": 11715
    },
    {
      "epoch": 0.8445181287392778,
      "grad_norm": 0.1165629038114653,
      "learning_rate": 1.241481117654687e-05,
      "loss": 1.0508,
      "step": 11716
    },
    {
      "epoch": 0.8445902112016146,
      "grad_norm": 0.10875457758188778,
      "learning_rate": 1.240354669522793e-05,
      "loss": 1.0482,
      "step": 11717
    },
    {
      "epoch": 0.8446622936639515,
      "grad_norm": 0.12185804795906671,
      "learning_rate": 1.2392286988699341e-05,
      "loss": 0.9533,
      "step": 11718
    },
    {
      "epoch": 0.8447343761262884,
      "grad_norm": 0.1216825476746743,
      "learning_rate": 1.2381032057574915e-05,
      "loss": 1.0032,
      "step": 11719
    },
    {
      "epoch": 0.8448064585886254,
      "grad_norm": 0.12035550597245198,
      "learning_rate": 1.236978190246808e-05,
      "loss": 1.0211,
      "step": 11720
    },
    {
      "epoch": 0.8448785410509623,
      "grad_norm": 0.1259891417929174,
      "learning_rate": 1.2358536523992103e-05,
      "loss": 1.0056,
      "step": 11721
    },
    {
      "epoch": 0.8449506235132992,
      "grad_norm": 0.11723251298499267,
      "learning_rate": 1.2347295922759938e-05,
      "loss": 1.0279,
      "step": 11722
    },
    {
      "epoch": 0.8450227059756361,
      "grad_norm": 0.22267995099818158,
      "learning_rate": 1.2336060099384306e-05,
      "loss": 1.034,
      "step": 11723
    },
    {
      "epoch": 0.845094788437973,
      "grad_norm": 0.15581003234129331,
      "learning_rate": 1.2324829054477671e-05,
      "loss": 1.0285,
      "step": 11724
    },
    {
      "epoch": 0.84516687090031,
      "grad_norm": 0.14122609464417316,
      "learning_rate": 1.2313602788652212e-05,
      "loss": 0.9661,
      "step": 11725
    },
    {
      "epoch": 0.8452389533626469,
      "grad_norm": 0.1259591231495163,
      "learning_rate": 1.2302381302519873e-05,
      "loss": 1.0367,
      "step": 11726
    },
    {
      "epoch": 0.8453110358249838,
      "grad_norm": 0.21333149262081771,
      "learning_rate": 1.2291164596692305e-05,
      "loss": 0.9795,
      "step": 11727
    },
    {
      "epoch": 0.8453831182873207,
      "grad_norm": 0.14185755297274769,
      "learning_rate": 1.2279952671780948e-05,
      "loss": 1.1032,
      "step": 11728
    },
    {
      "epoch": 0.8454552007496576,
      "grad_norm": 0.14440093460028663,
      "learning_rate": 1.226874552839693e-05,
      "loss": 1.0536,
      "step": 11729
    },
    {
      "epoch": 0.8455272832119946,
      "grad_norm": 0.11573160673654782,
      "learning_rate": 1.2257543167151142e-05,
      "loss": 1.0196,
      "step": 11730
    },
    {
      "epoch": 0.8455993656743315,
      "grad_norm": 0.13229459842244184,
      "learning_rate": 1.2246345588654218e-05,
      "loss": 0.9307,
      "step": 11731
    },
    {
      "epoch": 0.8456714481366684,
      "grad_norm": 0.12327824025135911,
      "learning_rate": 1.2235152793516536e-05,
      "loss": 1.0667,
      "step": 11732
    },
    {
      "epoch": 0.8457435305990053,
      "grad_norm": 0.16779046576799764,
      "learning_rate": 1.2223964782348185e-05,
      "loss": 1.0485,
      "step": 11733
    },
    {
      "epoch": 0.8458156130613422,
      "grad_norm": 0.12372747183526249,
      "learning_rate": 1.2212781555759022e-05,
      "loss": 1.0136,
      "step": 11734
    },
    {
      "epoch": 0.845887695523679,
      "grad_norm": 0.14595371416691066,
      "learning_rate": 1.2201603114358628e-05,
      "loss": 1.0743,
      "step": 11735
    },
    {
      "epoch": 0.845959777986016,
      "grad_norm": 0.13249580849490245,
      "learning_rate": 1.2190429458756336e-05,
      "loss": 1.0165,
      "step": 11736
    },
    {
      "epoch": 0.8460318604483529,
      "grad_norm": 0.11883498879892765,
      "learning_rate": 1.2179260589561204e-05,
      "loss": 0.986,
      "step": 11737
    },
    {
      "epoch": 0.8461039429106898,
      "grad_norm": 0.36691298772838654,
      "learning_rate": 1.2168096507382043e-05,
      "loss": 0.9543,
      "step": 11738
    },
    {
      "epoch": 0.8461760253730267,
      "grad_norm": 0.12858959320750402,
      "learning_rate": 1.2156937212827379e-05,
      "loss": 1.0638,
      "step": 11739
    },
    {
      "epoch": 0.8462481078353636,
      "grad_norm": 0.13688985180765367,
      "learning_rate": 1.2145782706505505e-05,
      "loss": 0.9662,
      "step": 11740
    },
    {
      "epoch": 0.8463201902977006,
      "grad_norm": 0.12487653086068375,
      "learning_rate": 1.2134632989024441e-05,
      "loss": 1.0356,
      "step": 11741
    },
    {
      "epoch": 0.8463922727600375,
      "grad_norm": 0.15194993666877452,
      "learning_rate": 1.212348806099195e-05,
      "loss": 1.0568,
      "step": 11742
    },
    {
      "epoch": 0.8464643552223744,
      "grad_norm": 0.12988734298005117,
      "learning_rate": 1.211234792301552e-05,
      "loss": 1.0096,
      "step": 11743
    },
    {
      "epoch": 0.8465364376847113,
      "grad_norm": 0.11440352780373927,
      "learning_rate": 1.2101212575702414e-05,
      "loss": 1.0781,
      "step": 11744
    },
    {
      "epoch": 0.8466085201470482,
      "grad_norm": 0.13110561435662307,
      "learning_rate": 1.2090082019659544e-05,
      "loss": 1.0721,
      "step": 11745
    },
    {
      "epoch": 0.8466806026093852,
      "grad_norm": 0.12242877405956995,
      "learning_rate": 1.2078956255493701e-05,
      "loss": 1.0721,
      "step": 11746
    },
    {
      "epoch": 0.8467526850717221,
      "grad_norm": 0.11263611561888555,
      "learning_rate": 1.206783528381129e-05,
      "loss": 1.0201,
      "step": 11747
    },
    {
      "epoch": 0.846824767534059,
      "grad_norm": 0.13210272043992072,
      "learning_rate": 1.2056719105218551e-05,
      "loss": 1.0626,
      "step": 11748
    },
    {
      "epoch": 0.8468968499963959,
      "grad_norm": 0.12038811555530449,
      "learning_rate": 1.204560772032134e-05,
      "loss": 1.0087,
      "step": 11749
    },
    {
      "epoch": 0.8469689324587328,
      "grad_norm": 0.12535982943034366,
      "learning_rate": 1.2034501129725396e-05,
      "loss": 1.0278,
      "step": 11750
    },
    {
      "epoch": 0.8470410149210698,
      "grad_norm": 0.1349935287217364,
      "learning_rate": 1.2023399334036089e-05,
      "loss": 1.0665,
      "step": 11751
    },
    {
      "epoch": 0.8471130973834066,
      "grad_norm": 0.12867310994314174,
      "learning_rate": 1.2012302333858616e-05,
      "loss": 1.0461,
      "step": 11752
    },
    {
      "epoch": 0.8471851798457435,
      "grad_norm": 0.11989114349429962,
      "learning_rate": 1.2001210129797779e-05,
      "loss": 1.0669,
      "step": 11753
    },
    {
      "epoch": 0.8472572623080804,
      "grad_norm": 0.12958998309709052,
      "learning_rate": 1.1990122722458264e-05,
      "loss": 0.9586,
      "step": 11754
    },
    {
      "epoch": 0.8473293447704173,
      "grad_norm": 0.15897602106643155,
      "learning_rate": 1.1979040112444428e-05,
      "loss": 1.0271,
      "step": 11755
    },
    {
      "epoch": 0.8474014272327542,
      "grad_norm": 0.12622720810600857,
      "learning_rate": 1.196796230036038e-05,
      "loss": 0.9851,
      "step": 11756
    },
    {
      "epoch": 0.8474735096950912,
      "grad_norm": 0.10930519684431182,
      "learning_rate": 1.1956889286809925e-05,
      "loss": 1.0412,
      "step": 11757
    },
    {
      "epoch": 0.8475455921574281,
      "grad_norm": 0.11858202126572182,
      "learning_rate": 1.1945821072396635e-05,
      "loss": 1.0289,
      "step": 11758
    },
    {
      "epoch": 0.847617674619765,
      "grad_norm": 0.11423805237803365,
      "learning_rate": 1.1934757657723882e-05,
      "loss": 1.0073,
      "step": 11759
    },
    {
      "epoch": 0.8476897570821019,
      "grad_norm": 0.11155670675754871,
      "learning_rate": 1.1923699043394699e-05,
      "loss": 1.0314,
      "step": 11760
    },
    {
      "epoch": 0.8477618395444388,
      "grad_norm": 0.11761180340079386,
      "learning_rate": 1.1912645230011854e-05,
      "loss": 1.0446,
      "step": 11761
    },
    {
      "epoch": 0.8478339220067758,
      "grad_norm": 0.13375458464069673,
      "learning_rate": 1.190159621817788e-05,
      "loss": 1.0216,
      "step": 11762
    },
    {
      "epoch": 0.8479060044691127,
      "grad_norm": 0.14563598572336647,
      "learning_rate": 1.1890552008495081e-05,
      "loss": 1.0254,
      "step": 11763
    },
    {
      "epoch": 0.8479780869314496,
      "grad_norm": 0.1565597138250181,
      "learning_rate": 1.1879512601565467e-05,
      "loss": 1.0569,
      "step": 11764
    },
    {
      "epoch": 0.8480501693937865,
      "grad_norm": 0.11846386549460007,
      "learning_rate": 1.1868477997990756e-05,
      "loss": 0.964,
      "step": 11765
    },
    {
      "epoch": 0.8481222518561234,
      "grad_norm": 0.1382514139680475,
      "learning_rate": 1.1857448198372423e-05,
      "loss": 1.0411,
      "step": 11766
    },
    {
      "epoch": 0.8481943343184604,
      "grad_norm": 0.1138930065903582,
      "learning_rate": 1.1846423203311696e-05,
      "loss": 1.0102,
      "step": 11767
    },
    {
      "epoch": 0.8482664167807973,
      "grad_norm": 0.17171923012280038,
      "learning_rate": 1.1835403013409596e-05,
      "loss": 0.9422,
      "step": 11768
    },
    {
      "epoch": 0.8483384992431341,
      "grad_norm": 0.14427706764350418,
      "learning_rate": 1.1824387629266742e-05,
      "loss": 1.0735,
      "step": 11769
    },
    {
      "epoch": 0.848410581705471,
      "grad_norm": 0.11990049531663355,
      "learning_rate": 1.1813377051483621e-05,
      "loss": 1.0006,
      "step": 11770
    },
    {
      "epoch": 0.8484826641678079,
      "grad_norm": 0.14506197175531368,
      "learning_rate": 1.1802371280660352e-05,
      "loss": 1.0136,
      "step": 11771
    },
    {
      "epoch": 0.8485547466301449,
      "grad_norm": 0.14342440768708767,
      "learning_rate": 1.1791370317396933e-05,
      "loss": 1.047,
      "step": 11772
    },
    {
      "epoch": 0.8486268290924818,
      "grad_norm": 0.11742432273056695,
      "learning_rate": 1.1780374162292951e-05,
      "loss": 0.9813,
      "step": 11773
    },
    {
      "epoch": 0.8486989115548187,
      "grad_norm": 0.12851223098654388,
      "learning_rate": 1.1769382815947816e-05,
      "loss": 0.9938,
      "step": 11774
    },
    {
      "epoch": 0.8487709940171556,
      "grad_norm": 0.328090050227142,
      "learning_rate": 1.1758396278960626e-05,
      "loss": 1.0666,
      "step": 11775
    },
    {
      "epoch": 0.8488430764794925,
      "grad_norm": 0.10340992937341825,
      "learning_rate": 1.1747414551930314e-05,
      "loss": 1.0347,
      "step": 11776
    },
    {
      "epoch": 0.8489151589418295,
      "grad_norm": 0.11715744707487309,
      "learning_rate": 1.1736437635455411e-05,
      "loss": 1.0685,
      "step": 11777
    },
    {
      "epoch": 0.8489872414041664,
      "grad_norm": 0.11997730662787251,
      "learning_rate": 1.1725465530134294e-05,
      "loss": 1.0486,
      "step": 11778
    },
    {
      "epoch": 0.8490593238665033,
      "grad_norm": 0.13729416383926396,
      "learning_rate": 1.171449823656503e-05,
      "loss": 0.978,
      "step": 11779
    },
    {
      "epoch": 0.8491314063288402,
      "grad_norm": 0.1366736396342378,
      "learning_rate": 1.170353575534544e-05,
      "loss": 1.0425,
      "step": 11780
    },
    {
      "epoch": 0.8492034887911771,
      "grad_norm": 0.11433825152952766,
      "learning_rate": 1.169257808707307e-05,
      "loss": 1.0094,
      "step": 11781
    },
    {
      "epoch": 0.849275571253514,
      "grad_norm": 0.10904226027117506,
      "learning_rate": 1.1681625232345217e-05,
      "loss": 0.9106,
      "step": 11782
    },
    {
      "epoch": 0.849347653715851,
      "grad_norm": 0.10882997414182892,
      "learning_rate": 1.1670677191758905e-05,
      "loss": 1.0938,
      "step": 11783
    },
    {
      "epoch": 0.8494197361781879,
      "grad_norm": 0.1096690447737673,
      "learning_rate": 1.1659733965910902e-05,
      "loss": 1.0293,
      "step": 11784
    },
    {
      "epoch": 0.8494918186405248,
      "grad_norm": 0.12613096931818712,
      "learning_rate": 1.1648795555397719e-05,
      "loss": 0.9814,
      "step": 11785
    },
    {
      "epoch": 0.8495639011028616,
      "grad_norm": 0.1960090742342328,
      "learning_rate": 1.1637861960815577e-05,
      "loss": 1.0541,
      "step": 11786
    },
    {
      "epoch": 0.8496359835651985,
      "grad_norm": 0.11296602924680363,
      "learning_rate": 1.162693318276048e-05,
      "loss": 0.9433,
      "step": 11787
    },
    {
      "epoch": 0.8497080660275355,
      "grad_norm": 0.16510048136628416,
      "learning_rate": 1.1616009221828129e-05,
      "loss": 1.0136,
      "step": 11788
    },
    {
      "epoch": 0.8497801484898724,
      "grad_norm": 0.13852112183517565,
      "learning_rate": 1.1605090078613978e-05,
      "loss": 1.0042,
      "step": 11789
    },
    {
      "epoch": 0.8498522309522093,
      "grad_norm": 0.101493067768786,
      "learning_rate": 1.1594175753713232e-05,
      "loss": 1.0807,
      "step": 11790
    },
    {
      "epoch": 0.8499243134145462,
      "grad_norm": 0.1290021120735238,
      "learning_rate": 1.1583266247720803e-05,
      "loss": 1.0694,
      "step": 11791
    },
    {
      "epoch": 0.8499963958768831,
      "grad_norm": 0.1307941150806173,
      "learning_rate": 1.157236156123136e-05,
      "loss": 1.0663,
      "step": 11792
    },
    {
      "epoch": 0.8500684783392201,
      "grad_norm": 0.12485810994735082,
      "learning_rate": 1.1561461694839303e-05,
      "loss": 0.9867,
      "step": 11793
    },
    {
      "epoch": 0.850140560801557,
      "grad_norm": 0.11944679671500072,
      "learning_rate": 1.1550566649138783e-05,
      "loss": 1.0377,
      "step": 11794
    },
    {
      "epoch": 0.8502126432638939,
      "grad_norm": 0.1314731764213324,
      "learning_rate": 1.1539676424723679e-05,
      "loss": 1.0237,
      "step": 11795
    },
    {
      "epoch": 0.8502847257262308,
      "grad_norm": 0.10826569733384134,
      "learning_rate": 1.152879102218759e-05,
      "loss": 1.023,
      "step": 11796
    },
    {
      "epoch": 0.8503568081885677,
      "grad_norm": 0.11977527653378338,
      "learning_rate": 1.1517910442123881e-05,
      "loss": 1.1474,
      "step": 11797
    },
    {
      "epoch": 0.8504288906509047,
      "grad_norm": 0.10720270052235795,
      "learning_rate": 1.1507034685125628e-05,
      "loss": 1.033,
      "step": 11798
    },
    {
      "epoch": 0.8505009731132416,
      "grad_norm": 0.2487935260969131,
      "learning_rate": 1.1496163751785672e-05,
      "loss": 1.063,
      "step": 11799
    },
    {
      "epoch": 0.8505730555755785,
      "grad_norm": 0.11700251626865715,
      "learning_rate": 1.1485297642696569e-05,
      "loss": 0.9741,
      "step": 11800
    },
    {
      "epoch": 0.8506451380379154,
      "grad_norm": 0.11441085670403328,
      "learning_rate": 1.1474436358450635e-05,
      "loss": 1.0118,
      "step": 11801
    },
    {
      "epoch": 0.8507172205002523,
      "grad_norm": 0.11527644517444925,
      "learning_rate": 1.1463579899639854e-05,
      "loss": 1.0646,
      "step": 11802
    },
    {
      "epoch": 0.8507893029625893,
      "grad_norm": 0.11221597250431593,
      "learning_rate": 1.1452728266856061e-05,
      "loss": 1.0482,
      "step": 11803
    },
    {
      "epoch": 0.8508613854249261,
      "grad_norm": 0.11168197525113929,
      "learning_rate": 1.1441881460690752e-05,
      "loss": 1.0842,
      "step": 11804
    },
    {
      "epoch": 0.850933467887263,
      "grad_norm": 0.12492550057128042,
      "learning_rate": 1.1431039481735174e-05,
      "loss": 1.0779,
      "step": 11805
    },
    {
      "epoch": 0.8510055503495999,
      "grad_norm": 0.15101525688635006,
      "learning_rate": 1.1420202330580266e-05,
      "loss": 1.0223,
      "step": 11806
    },
    {
      "epoch": 0.8510776328119368,
      "grad_norm": 0.11578122627166476,
      "learning_rate": 1.140937000781681e-05,
      "loss": 1.0173,
      "step": 11807
    },
    {
      "epoch": 0.8511497152742737,
      "grad_norm": 0.1366310367234561,
      "learning_rate": 1.1398542514035248e-05,
      "loss": 1.1043,
      "step": 11808
    },
    {
      "epoch": 0.8512217977366107,
      "grad_norm": 0.1316532862698938,
      "learning_rate": 1.1387719849825784e-05,
      "loss": 1.0634,
      "step": 11809
    },
    {
      "epoch": 0.8512938801989476,
      "grad_norm": 0.12852592842020275,
      "learning_rate": 1.1376902015778313e-05,
      "loss": 1.0598,
      "step": 11810
    },
    {
      "epoch": 0.8513659626612845,
      "grad_norm": 0.11993282978004775,
      "learning_rate": 1.1366089012482517e-05,
      "loss": 0.9605,
      "step": 11811
    },
    {
      "epoch": 0.8514380451236214,
      "grad_norm": 0.34937246769138414,
      "learning_rate": 1.135528084052785e-05,
      "loss": 1.0585,
      "step": 11812
    },
    {
      "epoch": 0.8515101275859583,
      "grad_norm": 0.15072918239972433,
      "learning_rate": 1.1344477500503391e-05,
      "loss": 1.0411,
      "step": 11813
    },
    {
      "epoch": 0.8515822100482953,
      "grad_norm": 0.11244693035894576,
      "learning_rate": 1.1333678992998043e-05,
      "loss": 1.0818,
      "step": 11814
    },
    {
      "epoch": 0.8516542925106322,
      "grad_norm": 0.10645142548160486,
      "learning_rate": 1.1322885318600407e-05,
      "loss": 1.0351,
      "step": 11815
    },
    {
      "epoch": 0.8517263749729691,
      "grad_norm": 0.10517440471965195,
      "learning_rate": 1.1312096477898881e-05,
      "loss": 1.038,
      "step": 11816
    },
    {
      "epoch": 0.851798457435306,
      "grad_norm": 0.12256291950426483,
      "learning_rate": 1.1301312471481507e-05,
      "loss": 1.0802,
      "step": 11817
    },
    {
      "epoch": 0.8518705398976429,
      "grad_norm": 0.11143437052178419,
      "learning_rate": 1.1290533299936123e-05,
      "loss": 0.9836,
      "step": 11818
    },
    {
      "epoch": 0.8519426223599799,
      "grad_norm": 0.10641185050386534,
      "learning_rate": 1.1279758963850273e-05,
      "loss": 0.9919,
      "step": 11819
    },
    {
      "epoch": 0.8520147048223168,
      "grad_norm": 0.13274553925354893,
      "learning_rate": 1.1268989463811308e-05,
      "loss": 1.061,
      "step": 11820
    },
    {
      "epoch": 0.8520867872846536,
      "grad_norm": 0.10900691259691202,
      "learning_rate": 1.1258224800406214e-05,
      "loss": 0.996,
      "step": 11821
    },
    {
      "epoch": 0.8521588697469905,
      "grad_norm": 0.11115249106319026,
      "learning_rate": 1.1247464974221778e-05,
      "loss": 1.0122,
      "step": 11822
    },
    {
      "epoch": 0.8522309522093274,
      "grad_norm": 0.12559357725942097,
      "learning_rate": 1.1236709985844497e-05,
      "loss": 1.023,
      "step": 11823
    },
    {
      "epoch": 0.8523030346716644,
      "grad_norm": 0.11439902133667673,
      "learning_rate": 1.1225959835860622e-05,
      "loss": 0.9674,
      "step": 11824
    },
    {
      "epoch": 0.8523751171340013,
      "grad_norm": 0.13176573175795717,
      "learning_rate": 1.1215214524856133e-05,
      "loss": 1.0299,
      "step": 11825
    },
    {
      "epoch": 0.8524471995963382,
      "grad_norm": 0.15162178880585928,
      "learning_rate": 1.1204474053416737e-05,
      "loss": 1.0524,
      "step": 11826
    },
    {
      "epoch": 0.8525192820586751,
      "grad_norm": 0.1437377574506283,
      "learning_rate": 1.1193738422127897e-05,
      "loss": 0.967,
      "step": 11827
    },
    {
      "epoch": 0.852591364521012,
      "grad_norm": 0.12042968654018373,
      "learning_rate": 1.1183007631574805e-05,
      "loss": 1.0702,
      "step": 11828
    },
    {
      "epoch": 0.852663446983349,
      "grad_norm": 0.12193477582380405,
      "learning_rate": 1.1172281682342367e-05,
      "loss": 1.0328,
      "step": 11829
    },
    {
      "epoch": 0.8527355294456859,
      "grad_norm": 0.11229543441702043,
      "learning_rate": 1.116156057501525e-05,
      "loss": 0.9881,
      "step": 11830
    },
    {
      "epoch": 0.8528076119080228,
      "grad_norm": 0.12108557739166648,
      "learning_rate": 1.1150844310177865e-05,
      "loss": 1.0046,
      "step": 11831
    },
    {
      "epoch": 0.8528796943703597,
      "grad_norm": 0.1302804950806728,
      "learning_rate": 1.1140132888414323e-05,
      "loss": 1.1187,
      "step": 11832
    },
    {
      "epoch": 0.8529517768326966,
      "grad_norm": 0.1269748340114511,
      "learning_rate": 1.11294263103085e-05,
      "loss": 1.0465,
      "step": 11833
    },
    {
      "epoch": 0.8530238592950335,
      "grad_norm": 0.12328747409479593,
      "learning_rate": 1.1118724576444006e-05,
      "loss": 0.9659,
      "step": 11834
    },
    {
      "epoch": 0.8530959417573705,
      "grad_norm": 0.1199383538425101,
      "learning_rate": 1.1108027687404188e-05,
      "loss": 1.0457,
      "step": 11835
    },
    {
      "epoch": 0.8531680242197074,
      "grad_norm": 0.1210568000218153,
      "learning_rate": 1.10973356437721e-05,
      "loss": 1.0252,
      "step": 11836
    },
    {
      "epoch": 0.8532401066820443,
      "grad_norm": 0.12910778880517143,
      "learning_rate": 1.1086648446130566e-05,
      "loss": 0.9832,
      "step": 11837
    },
    {
      "epoch": 0.8533121891443811,
      "grad_norm": 0.17102238053749322,
      "learning_rate": 1.1075966095062129e-05,
      "loss": 1.002,
      "step": 11838
    },
    {
      "epoch": 0.853384271606718,
      "grad_norm": 0.14436742522505672,
      "learning_rate": 1.106528859114908e-05,
      "loss": 1.0117,
      "step": 11839
    },
    {
      "epoch": 0.853456354069055,
      "grad_norm": 0.10869211593559137,
      "learning_rate": 1.1054615934973433e-05,
      "loss": 1.0017,
      "step": 11840
    },
    {
      "epoch": 0.8535284365313919,
      "grad_norm": 0.11574636479667079,
      "learning_rate": 1.1043948127116954e-05,
      "loss": 1.0586,
      "step": 11841
    },
    {
      "epoch": 0.8536005189937288,
      "grad_norm": 0.12750446044082633,
      "learning_rate": 1.1033285168161123e-05,
      "loss": 1.0834,
      "step": 11842
    },
    {
      "epoch": 0.8536726014560657,
      "grad_norm": 0.11548595473290572,
      "learning_rate": 1.1022627058687163e-05,
      "loss": 1.0345,
      "step": 11843
    },
    {
      "epoch": 0.8537446839184026,
      "grad_norm": 0.12682017197377823,
      "learning_rate": 1.1011973799276054e-05,
      "loss": 0.9795,
      "step": 11844
    },
    {
      "epoch": 0.8538167663807396,
      "grad_norm": 0.11812600289163629,
      "learning_rate": 1.1001325390508488e-05,
      "loss": 1.1216,
      "step": 11845
    },
    {
      "epoch": 0.8538888488430765,
      "grad_norm": 0.11605944362378798,
      "learning_rate": 1.0990681832964866e-05,
      "loss": 1.0214,
      "step": 11846
    },
    {
      "epoch": 0.8539609313054134,
      "grad_norm": 0.12469271849635742,
      "learning_rate": 1.0980043127225393e-05,
      "loss": 1.0608,
      "step": 11847
    },
    {
      "epoch": 0.8540330137677503,
      "grad_norm": 0.11437535025087392,
      "learning_rate": 1.0969409273869979e-05,
      "loss": 1.0378,
      "step": 11848
    },
    {
      "epoch": 0.8541050962300872,
      "grad_norm": 0.1222329140207839,
      "learning_rate": 1.0958780273478253e-05,
      "loss": 1.0165,
      "step": 11849
    },
    {
      "epoch": 0.8541771786924242,
      "grad_norm": 0.1266073151238848,
      "learning_rate": 1.0948156126629561e-05,
      "loss": 1.0573,
      "step": 11850
    },
    {
      "epoch": 0.8542492611547611,
      "grad_norm": 0.10726291348877391,
      "learning_rate": 1.093753683390305e-05,
      "loss": 0.9965,
      "step": 11851
    },
    {
      "epoch": 0.854321343617098,
      "grad_norm": 0.1388926054452939,
      "learning_rate": 1.092692239587757e-05,
      "loss": 1.0027,
      "step": 11852
    },
    {
      "epoch": 0.8543934260794349,
      "grad_norm": 0.12521274350961478,
      "learning_rate": 1.0916312813131702e-05,
      "loss": 1.039,
      "step": 11853
    },
    {
      "epoch": 0.8544655085417718,
      "grad_norm": 0.36172096974656665,
      "learning_rate": 1.0905708086243727e-05,
      "loss": 1.0194,
      "step": 11854
    },
    {
      "epoch": 0.8545375910041088,
      "grad_norm": 0.14206137184647058,
      "learning_rate": 1.0895108215791717e-05,
      "loss": 1.0532,
      "step": 11855
    },
    {
      "epoch": 0.8546096734664456,
      "grad_norm": 0.12239347302948939,
      "learning_rate": 1.0884513202353463e-05,
      "loss": 1.1086,
      "step": 11856
    },
    {
      "epoch": 0.8546817559287825,
      "grad_norm": 0.10592474240473636,
      "learning_rate": 1.087392304650653e-05,
      "loss": 1.0563,
      "step": 11857
    },
    {
      "epoch": 0.8547538383911194,
      "grad_norm": 0.16253107477969014,
      "learning_rate": 1.0863337748828106e-05,
      "loss": 1.0777,
      "step": 11858
    },
    {
      "epoch": 0.8548259208534563,
      "grad_norm": 0.12279803378248064,
      "learning_rate": 1.0852757309895189e-05,
      "loss": 1.0247,
      "step": 11859
    },
    {
      "epoch": 0.8548980033157932,
      "grad_norm": 0.12584305175612517,
      "learning_rate": 1.0842181730284562e-05,
      "loss": 0.9261,
      "step": 11860
    },
    {
      "epoch": 0.8549700857781302,
      "grad_norm": 0.12028673874034414,
      "learning_rate": 1.0831611010572674e-05,
      "loss": 1.076,
      "step": 11861
    },
    {
      "epoch": 0.8550421682404671,
      "grad_norm": 0.12584902031198025,
      "learning_rate": 1.0821045151335685e-05,
      "loss": 1.0601,
      "step": 11862
    },
    {
      "epoch": 0.855114250702804,
      "grad_norm": 0.11357901262366638,
      "learning_rate": 1.0810484153149548e-05,
      "loss": 1.061,
      "step": 11863
    },
    {
      "epoch": 0.8551863331651409,
      "grad_norm": 0.13700790957991046,
      "learning_rate": 1.0799928016589945e-05,
      "loss": 0.9782,
      "step": 11864
    },
    {
      "epoch": 0.8552584156274778,
      "grad_norm": 0.11971919136104885,
      "learning_rate": 1.0789376742232293e-05,
      "loss": 1.0736,
      "step": 11865
    },
    {
      "epoch": 0.8553304980898148,
      "grad_norm": 0.11503742351111118,
      "learning_rate": 1.077883033065169e-05,
      "loss": 1.0787,
      "step": 11866
    },
    {
      "epoch": 0.8554025805521517,
      "grad_norm": 0.12787624520234891,
      "learning_rate": 1.0768288782423031e-05,
      "loss": 1.064,
      "step": 11867
    },
    {
      "epoch": 0.8554746630144886,
      "grad_norm": 0.12364224745442781,
      "learning_rate": 1.0757752098120899e-05,
      "loss": 0.9809,
      "step": 11868
    },
    {
      "epoch": 0.8555467454768255,
      "grad_norm": 0.1212954096746375,
      "learning_rate": 1.0747220278319703e-05,
      "loss": 0.992,
      "step": 11869
    },
    {
      "epoch": 0.8556188279391624,
      "grad_norm": 0.15082226959209477,
      "learning_rate": 1.0736693323593461e-05,
      "loss": 1.0455,
      "step": 11870
    },
    {
      "epoch": 0.8556909104014994,
      "grad_norm": 0.12029812258594551,
      "learning_rate": 1.0726171234516015e-05,
      "loss": 1.0555,
      "step": 11871
    },
    {
      "epoch": 0.8557629928638363,
      "grad_norm": 0.15000467432430373,
      "learning_rate": 1.071565401166087e-05,
      "loss": 1.0043,
      "step": 11872
    },
    {
      "epoch": 0.8558350753261731,
      "grad_norm": 0.1094963398548918,
      "learning_rate": 1.0705141655601381e-05,
      "loss": 1.004,
      "step": 11873
    },
    {
      "epoch": 0.85590715778851,
      "grad_norm": 0.1086767235731311,
      "learning_rate": 1.0694634166910511e-05,
      "loss": 0.9966,
      "step": 11874
    },
    {
      "epoch": 0.8559792402508469,
      "grad_norm": 0.35748891282952205,
      "learning_rate": 1.0684131546161035e-05,
      "loss": 1.0512,
      "step": 11875
    },
    {
      "epoch": 0.8560513227131838,
      "grad_norm": 0.11311175138137508,
      "learning_rate": 1.0673633793925419e-05,
      "loss": 1.0039,
      "step": 11876
    },
    {
      "epoch": 0.8561234051755208,
      "grad_norm": 0.13165144197172915,
      "learning_rate": 1.0663140910775904e-05,
      "loss": 1.0245,
      "step": 11877
    },
    {
      "epoch": 0.8561954876378577,
      "grad_norm": 1.062155662463262,
      "learning_rate": 1.0652652897284443e-05,
      "loss": 0.9379,
      "step": 11878
    },
    {
      "epoch": 0.8562675701001946,
      "grad_norm": 0.12394801607026336,
      "learning_rate": 1.0642169754022724e-05,
      "loss": 0.9444,
      "step": 11879
    },
    {
      "epoch": 0.8563396525625315,
      "grad_norm": 0.12189293460361109,
      "learning_rate": 1.0631691481562167e-05,
      "loss": 1.0203,
      "step": 11880
    },
    {
      "epoch": 0.8564117350248684,
      "grad_norm": 0.12283409127363026,
      "learning_rate": 1.0621218080473927e-05,
      "loss": 1.0431,
      "step": 11881
    },
    {
      "epoch": 0.8564838174872054,
      "grad_norm": 0.12279561103659999,
      "learning_rate": 1.0610749551328925e-05,
      "loss": 1.0106,
      "step": 11882
    },
    {
      "epoch": 0.8565558999495423,
      "grad_norm": 0.12172417460555375,
      "learning_rate": 1.060028589469776e-05,
      "loss": 0.9907,
      "step": 11883
    },
    {
      "epoch": 0.8566279824118792,
      "grad_norm": 0.1245026949094479,
      "learning_rate": 1.0589827111150808e-05,
      "loss": 0.965,
      "step": 11884
    },
    {
      "epoch": 0.8567000648742161,
      "grad_norm": 0.1368859276394142,
      "learning_rate": 1.057937320125817e-05,
      "loss": 1.0688,
      "step": 11885
    },
    {
      "epoch": 0.856772147336553,
      "grad_norm": 0.13365825517352567,
      "learning_rate": 1.0568924165589666e-05,
      "loss": 1.0326,
      "step": 11886
    },
    {
      "epoch": 0.85684422979889,
      "grad_norm": 0.12298016932747631,
      "learning_rate": 1.0558480004714866e-05,
      "loss": 1.0675,
      "step": 11887
    },
    {
      "epoch": 0.8569163122612269,
      "grad_norm": 0.11087164157820037,
      "learning_rate": 1.054804071920308e-05,
      "loss": 1.0416,
      "step": 11888
    },
    {
      "epoch": 0.8569883947235638,
      "grad_norm": 0.12167180252905063,
      "learning_rate": 1.0537606309623338e-05,
      "loss": 0.9719,
      "step": 11889
    },
    {
      "epoch": 0.8570604771859006,
      "grad_norm": 0.14911547761654545,
      "learning_rate": 1.0527176776544389e-05,
      "loss": 0.9243,
      "step": 11890
    },
    {
      "epoch": 0.8571325596482375,
      "grad_norm": 0.12570983820073886,
      "learning_rate": 1.0516752120534768e-05,
      "loss": 1.003,
      "step": 11891
    },
    {
      "epoch": 0.8572046421105745,
      "grad_norm": 0.17238546068172425,
      "learning_rate": 1.0506332342162695e-05,
      "loss": 1.0804,
      "step": 11892
    },
    {
      "epoch": 0.8572767245729114,
      "grad_norm": 0.22207595434459734,
      "learning_rate": 1.049591744199614e-05,
      "loss": 1.0287,
      "step": 11893
    },
    {
      "epoch": 0.8573488070352483,
      "grad_norm": 0.17551801010147747,
      "learning_rate": 1.0485507420602803e-05,
      "loss": 1.074,
      "step": 11894
    },
    {
      "epoch": 0.8574208894975852,
      "grad_norm": 0.11795250340659093,
      "learning_rate": 1.0475102278550142e-05,
      "loss": 1.0192,
      "step": 11895
    },
    {
      "epoch": 0.8574929719599221,
      "grad_norm": 0.1259669471790042,
      "learning_rate": 1.0464702016405314e-05,
      "loss": 1.0233,
      "step": 11896
    },
    {
      "epoch": 0.857565054422259,
      "grad_norm": 0.12916055824942446,
      "learning_rate": 1.0454306634735234e-05,
      "loss": 1.0409,
      "step": 11897
    },
    {
      "epoch": 0.857637136884596,
      "grad_norm": 0.15654190834598025,
      "learning_rate": 1.0443916134106558e-05,
      "loss": 1.0208,
      "step": 11898
    },
    {
      "epoch": 0.8577092193469329,
      "grad_norm": 0.13260544627222468,
      "learning_rate": 1.043353051508561e-05,
      "loss": 1.0171,
      "step": 11899
    },
    {
      "epoch": 0.8577813018092698,
      "grad_norm": 0.141623500664096,
      "learning_rate": 1.042314977823855e-05,
      "loss": 1.0173,
      "step": 11900
    },
    {
      "epoch": 0.8578533842716067,
      "grad_norm": 0.16421593434487983,
      "learning_rate": 1.0412773924131203e-05,
      "loss": 1.017,
      "step": 11901
    },
    {
      "epoch": 0.8579254667339437,
      "grad_norm": 0.12252861446524112,
      "learning_rate": 1.0402402953329171e-05,
      "loss": 1.0584,
      "step": 11902
    },
    {
      "epoch": 0.8579975491962806,
      "grad_norm": 0.12401333403022148,
      "learning_rate": 1.0392036866397703e-05,
      "loss": 1.0441,
      "step": 11903
    },
    {
      "epoch": 0.8580696316586175,
      "grad_norm": 0.12616302366549162,
      "learning_rate": 1.038167566390189e-05,
      "loss": 1.0291,
      "step": 11904
    },
    {
      "epoch": 0.8581417141209544,
      "grad_norm": 0.12310497792952954,
      "learning_rate": 1.0371319346406516e-05,
      "loss": 1.0869,
      "step": 11905
    },
    {
      "epoch": 0.8582137965832913,
      "grad_norm": 0.20534774868904676,
      "learning_rate": 1.0360967914476083e-05,
      "loss": 1.0212,
      "step": 11906
    },
    {
      "epoch": 0.8582858790456283,
      "grad_norm": 0.11808542235508868,
      "learning_rate": 1.0350621368674807e-05,
      "loss": 1.0414,
      "step": 11907
    },
    {
      "epoch": 0.8583579615079651,
      "grad_norm": 0.1328191407246857,
      "learning_rate": 1.0340279709566702e-05,
      "loss": 0.9715,
      "step": 11908
    },
    {
      "epoch": 0.858430043970302,
      "grad_norm": 0.15559407838712144,
      "learning_rate": 1.0329942937715475e-05,
      "loss": 0.9677,
      "step": 11909
    },
    {
      "epoch": 0.8585021264326389,
      "grad_norm": 0.1477498000664855,
      "learning_rate": 1.0319611053684596e-05,
      "loss": 1.034,
      "step": 11910
    },
    {
      "epoch": 0.8585742088949758,
      "grad_norm": 0.11638021465323288,
      "learning_rate": 1.0309284058037194e-05,
      "loss": 0.988,
      "step": 11911
    },
    {
      "epoch": 0.8586462913573127,
      "grad_norm": 0.12103514421482225,
      "learning_rate": 1.0298961951336184e-05,
      "loss": 1.0441,
      "step": 11912
    },
    {
      "epoch": 0.8587183738196497,
      "grad_norm": 0.12468635250490298,
      "learning_rate": 1.0288644734144259e-05,
      "loss": 1.0403,
      "step": 11913
    },
    {
      "epoch": 0.8587904562819866,
      "grad_norm": 0.14456985461188093,
      "learning_rate": 1.0278332407023794e-05,
      "loss": 1.0166,
      "step": 11914
    },
    {
      "epoch": 0.8588625387443235,
      "grad_norm": 0.1156294049742342,
      "learning_rate": 1.0268024970536871e-05,
      "loss": 1.0161,
      "step": 11915
    },
    {
      "epoch": 0.8589346212066604,
      "grad_norm": 0.19875115371192348,
      "learning_rate": 1.025772242524533e-05,
      "loss": 1.0442,
      "step": 11916
    },
    {
      "epoch": 0.8590067036689973,
      "grad_norm": 0.11661572775828534,
      "learning_rate": 1.024742477171079e-05,
      "loss": 1.0468,
      "step": 11917
    },
    {
      "epoch": 0.8590787861313343,
      "grad_norm": 0.15901872289232474,
      "learning_rate": 1.0237132010494577e-05,
      "loss": 1.0719,
      "step": 11918
    },
    {
      "epoch": 0.8591508685936712,
      "grad_norm": 0.12196204038316114,
      "learning_rate": 1.0226844142157677e-05,
      "loss": 1.0296,
      "step": 11919
    },
    {
      "epoch": 0.8592229510560081,
      "grad_norm": 0.11905731164162144,
      "learning_rate": 1.021656116726092e-05,
      "loss": 1.1016,
      "step": 11920
    },
    {
      "epoch": 0.859295033518345,
      "grad_norm": 0.11844584483720941,
      "learning_rate": 1.0206283086364776e-05,
      "loss": 1.0083,
      "step": 11921
    },
    {
      "epoch": 0.8593671159806819,
      "grad_norm": 0.12419669297859026,
      "learning_rate": 1.0196009900029568e-05,
      "loss": 1.1029,
      "step": 11922
    },
    {
      "epoch": 0.8594391984430189,
      "grad_norm": 0.12810036093376695,
      "learning_rate": 1.0185741608815202e-05,
      "loss": 1.0607,
      "step": 11923
    },
    {
      "epoch": 0.8595112809053558,
      "grad_norm": 0.16830117951103285,
      "learning_rate": 1.017547821328143e-05,
      "loss": 1.0063,
      "step": 11924
    },
    {
      "epoch": 0.8595833633676926,
      "grad_norm": 0.13394378140615604,
      "learning_rate": 1.0165219713987662e-05,
      "loss": 1.0271,
      "step": 11925
    },
    {
      "epoch": 0.8596554458300295,
      "grad_norm": 0.12991029101002186,
      "learning_rate": 1.0154966111493147e-05,
      "loss": 1.1023,
      "step": 11926
    },
    {
      "epoch": 0.8597275282923664,
      "grad_norm": 0.17334245170678544,
      "learning_rate": 1.0144717406356729e-05,
      "loss": 1.0875,
      "step": 11927
    },
    {
      "epoch": 0.8597996107547033,
      "grad_norm": 0.12228491386843394,
      "learning_rate": 1.0134473599137084e-05,
      "loss": 1.0402,
      "step": 11928
    },
    {
      "epoch": 0.8598716932170403,
      "grad_norm": 0.13818523031980506,
      "learning_rate": 1.0124234690392586e-05,
      "loss": 1.1043,
      "step": 11929
    },
    {
      "epoch": 0.8599437756793772,
      "grad_norm": 0.12496090776323228,
      "learning_rate": 1.0114000680681357e-05,
      "loss": 0.9379,
      "step": 11930
    },
    {
      "epoch": 0.8600158581417141,
      "grad_norm": 0.1416183696438862,
      "learning_rate": 1.0103771570561226e-05,
      "loss": 0.98,
      "step": 11931
    },
    {
      "epoch": 0.860087940604051,
      "grad_norm": 0.13668679145015322,
      "learning_rate": 1.0093547360589784e-05,
      "loss": 1.1014,
      "step": 11932
    },
    {
      "epoch": 0.860160023066388,
      "grad_norm": 0.23385559079917254,
      "learning_rate": 1.0083328051324337e-05,
      "loss": 1.0485,
      "step": 11933
    },
    {
      "epoch": 0.8602321055287249,
      "grad_norm": 0.10322087871454215,
      "learning_rate": 1.0073113643321929e-05,
      "loss": 1.0723,
      "step": 11934
    },
    {
      "epoch": 0.8603041879910618,
      "grad_norm": 0.11389376211462193,
      "learning_rate": 1.006290413713933e-05,
      "loss": 1.0567,
      "step": 11935
    },
    {
      "epoch": 0.8603762704533987,
      "grad_norm": 0.1343057103507227,
      "learning_rate": 1.0052699533333065e-05,
      "loss": 1.0449,
      "step": 11936
    },
    {
      "epoch": 0.8604483529157356,
      "grad_norm": 0.14144944901836287,
      "learning_rate": 1.0042499832459362e-05,
      "loss": 1.0556,
      "step": 11937
    },
    {
      "epoch": 0.8605204353780725,
      "grad_norm": 0.12293142532880019,
      "learning_rate": 1.0032305035074197e-05,
      "loss": 1.0482,
      "step": 11938
    },
    {
      "epoch": 0.8605925178404095,
      "grad_norm": 0.13039489284884856,
      "learning_rate": 1.0022115141733279e-05,
      "loss": 1.0341,
      "step": 11939
    },
    {
      "epoch": 0.8606646003027464,
      "grad_norm": 0.12716146289850383,
      "learning_rate": 1.0011930152992067e-05,
      "loss": 1.0322,
      "step": 11940
    },
    {
      "epoch": 0.8607366827650833,
      "grad_norm": 0.11975281818320117,
      "learning_rate": 1.0001750069405701e-05,
      "loss": 0.9792,
      "step": 11941
    },
    {
      "epoch": 0.8608087652274201,
      "grad_norm": 0.12130550086274604,
      "learning_rate": 9.991574891529109e-06,
      "loss": 1.0356,
      "step": 11942
    },
    {
      "epoch": 0.860880847689757,
      "grad_norm": 0.13571163799582986,
      "learning_rate": 9.981404619916934e-06,
      "loss": 1.1158,
      "step": 11943
    },
    {
      "epoch": 0.860952930152094,
      "grad_norm": 0.11233286885954599,
      "learning_rate": 9.97123925512352e-06,
      "loss": 1.0255,
      "step": 11944
    },
    {
      "epoch": 0.8610250126144309,
      "grad_norm": 0.1177836312273108,
      "learning_rate": 9.961078797703005e-06,
      "loss": 0.9731,
      "step": 11945
    },
    {
      "epoch": 0.8610970950767678,
      "grad_norm": 0.12278108450925708,
      "learning_rate": 9.950923248209188e-06,
      "loss": 1.0919,
      "step": 11946
    },
    {
      "epoch": 0.8611691775391047,
      "grad_norm": 0.11163778243798306,
      "learning_rate": 9.94077260719567e-06,
      "loss": 1.099,
      "step": 11947
    },
    {
      "epoch": 0.8612412600014416,
      "grad_norm": 0.13197987442397782,
      "learning_rate": 9.930626875215731e-06,
      "loss": 1.0814,
      "step": 11948
    },
    {
      "epoch": 0.8613133424637786,
      "grad_norm": 0.1390388734240552,
      "learning_rate": 9.920486052822409e-06,
      "loss": 0.9665,
      "step": 11949
    },
    {
      "epoch": 0.8613854249261155,
      "grad_norm": 0.12474963826796592,
      "learning_rate": 9.910350140568481e-06,
      "loss": 0.9359,
      "step": 11950
    },
    {
      "epoch": 0.8614575073884524,
      "grad_norm": 0.14502372500165736,
      "learning_rate": 9.90021913900645e-06,
      "loss": 1.1031,
      "step": 11951
    },
    {
      "epoch": 0.8615295898507893,
      "grad_norm": 0.1303129943332835,
      "learning_rate": 9.890093048688487e-06,
      "loss": 1.0237,
      "step": 11952
    },
    {
      "epoch": 0.8616016723131262,
      "grad_norm": 0.12275255300499788,
      "learning_rate": 9.879971870166626e-06,
      "loss": 1.0414,
      "step": 11953
    },
    {
      "epoch": 0.8616737547754632,
      "grad_norm": 0.7445226917848993,
      "learning_rate": 9.869855603992529e-06,
      "loss": 0.9821,
      "step": 11954
    },
    {
      "epoch": 0.8617458372378001,
      "grad_norm": 0.1456874653352041,
      "learning_rate": 9.859744250717651e-06,
      "loss": 1.0899,
      "step": 11955
    },
    {
      "epoch": 0.861817919700137,
      "grad_norm": 0.1229896475274499,
      "learning_rate": 9.849637810893098e-06,
      "loss": 0.9932,
      "step": 11956
    },
    {
      "epoch": 0.8618900021624739,
      "grad_norm": 0.12392830044263299,
      "learning_rate": 9.839536285069805e-06,
      "loss": 1.0447,
      "step": 11957
    },
    {
      "epoch": 0.8619620846248108,
      "grad_norm": 0.11373063774090718,
      "learning_rate": 9.82943967379838e-06,
      "loss": 0.9638,
      "step": 11958
    },
    {
      "epoch": 0.8620341670871478,
      "grad_norm": 0.11561345807872628,
      "learning_rate": 9.819347977629202e-06,
      "loss": 1.0375,
      "step": 11959
    },
    {
      "epoch": 0.8621062495494846,
      "grad_norm": 0.11145684854810621,
      "learning_rate": 9.809261197112296e-06,
      "loss": 1.0366,
      "step": 11960
    },
    {
      "epoch": 0.8621783320118215,
      "grad_norm": 0.13146439266419407,
      "learning_rate": 9.79917933279756e-06,
      "loss": 0.9955,
      "step": 11961
    },
    {
      "epoch": 0.8622504144741584,
      "grad_norm": 0.11954141328304668,
      "learning_rate": 9.789102385234516e-06,
      "loss": 1.03,
      "step": 11962
    },
    {
      "epoch": 0.8623224969364953,
      "grad_norm": 0.12443098080738328,
      "learning_rate": 9.779030354972419e-06,
      "loss": 0.9909,
      "step": 11963
    },
    {
      "epoch": 0.8623945793988322,
      "grad_norm": 0.1294351306659731,
      "learning_rate": 9.768963242560313e-06,
      "loss": 1.1167,
      "step": 11964
    },
    {
      "epoch": 0.8624666618611692,
      "grad_norm": 0.1188137642886964,
      "learning_rate": 9.75890104854692e-06,
      "loss": 1.0071,
      "step": 11965
    },
    {
      "epoch": 0.8625387443235061,
      "grad_norm": 0.10882772288288045,
      "learning_rate": 9.748843773480787e-06,
      "loss": 0.9927,
      "step": 11966
    },
    {
      "epoch": 0.862610826785843,
      "grad_norm": 0.14091276644440098,
      "learning_rate": 9.738791417910043e-06,
      "loss": 1.1339,
      "step": 11967
    },
    {
      "epoch": 0.8626829092481799,
      "grad_norm": 0.11107472739597231,
      "learning_rate": 9.728743982382681e-06,
      "loss": 1.0178,
      "step": 11968
    },
    {
      "epoch": 0.8627549917105168,
      "grad_norm": 0.22248981464818937,
      "learning_rate": 9.718701467446333e-06,
      "loss": 1.0545,
      "step": 11969
    },
    {
      "epoch": 0.8628270741728538,
      "grad_norm": 0.11832636279728301,
      "learning_rate": 9.708663873648482e-06,
      "loss": 1.0819,
      "step": 11970
    },
    {
      "epoch": 0.8628991566351907,
      "grad_norm": 0.15454211342704421,
      "learning_rate": 9.69863120153619e-06,
      "loss": 1.0904,
      "step": 11971
    },
    {
      "epoch": 0.8629712390975276,
      "grad_norm": 0.1313832088767309,
      "learning_rate": 9.688603451656364e-06,
      "loss": 1.0866,
      "step": 11972
    },
    {
      "epoch": 0.8630433215598645,
      "grad_norm": 0.6869580125710496,
      "learning_rate": 9.67858062455559e-06,
      "loss": 1.0327,
      "step": 11973
    },
    {
      "epoch": 0.8631154040222014,
      "grad_norm": 0.36182761346189773,
      "learning_rate": 9.668562720780217e-06,
      "loss": 1.0579,
      "step": 11974
    },
    {
      "epoch": 0.8631874864845384,
      "grad_norm": 0.10764214991039738,
      "learning_rate": 9.6585497408763e-06,
      "loss": 1.0578,
      "step": 11975
    },
    {
      "epoch": 0.8632595689468753,
      "grad_norm": 0.12759680270470058,
      "learning_rate": 9.648541685389645e-06,
      "loss": 1.1031,
      "step": 11976
    },
    {
      "epoch": 0.8633316514092121,
      "grad_norm": 0.13928069406189608,
      "learning_rate": 9.638538554865773e-06,
      "loss": 1.0359,
      "step": 11977
    },
    {
      "epoch": 0.863403733871549,
      "grad_norm": 0.12919982524017903,
      "learning_rate": 9.628540349849946e-06,
      "loss": 1.0152,
      "step": 11978
    },
    {
      "epoch": 0.8634758163338859,
      "grad_norm": 0.14147825095509303,
      "learning_rate": 9.618547070887162e-06,
      "loss": 1.0205,
      "step": 11979
    },
    {
      "epoch": 0.8635478987962228,
      "grad_norm": 0.12489821014952412,
      "learning_rate": 9.608558718522132e-06,
      "loss": 1.0298,
      "step": 11980
    },
    {
      "epoch": 0.8636199812585598,
      "grad_norm": 0.1705767320487547,
      "learning_rate": 9.598575293299317e-06,
      "loss": 1.0533,
      "step": 11981
    },
    {
      "epoch": 0.8636920637208967,
      "grad_norm": 0.11266605583567194,
      "learning_rate": 9.588596795762906e-06,
      "loss": 1.0206,
      "step": 11982
    },
    {
      "epoch": 0.8637641461832336,
      "grad_norm": 0.11147582855046036,
      "learning_rate": 9.578623226456806e-06,
      "loss": 0.9274,
      "step": 11983
    },
    {
      "epoch": 0.8638362286455705,
      "grad_norm": 0.13217387365184613,
      "learning_rate": 9.568654585924685e-06,
      "loss": 0.9856,
      "step": 11984
    },
    {
      "epoch": 0.8639083111079074,
      "grad_norm": 0.12472720845365463,
      "learning_rate": 9.558690874709898e-06,
      "loss": 0.9905,
      "step": 11985
    },
    {
      "epoch": 0.8639803935702444,
      "grad_norm": 0.12350861251230075,
      "learning_rate": 9.548732093355572e-06,
      "loss": 0.9584,
      "step": 11986
    },
    {
      "epoch": 0.8640524760325813,
      "grad_norm": 0.1667842576068744,
      "learning_rate": 9.538778242404555e-06,
      "loss": 0.9865,
      "step": 11987
    },
    {
      "epoch": 0.8641245584949182,
      "grad_norm": 0.11401016642173871,
      "learning_rate": 9.528829322399414e-06,
      "loss": 1.0297,
      "step": 11988
    },
    {
      "epoch": 0.8641966409572551,
      "grad_norm": 0.1195834467660055,
      "learning_rate": 9.518885333882443e-06,
      "loss": 0.9785,
      "step": 11989
    },
    {
      "epoch": 0.864268723419592,
      "grad_norm": 0.11127999217025193,
      "learning_rate": 9.508946277395691e-06,
      "loss": 0.9923,
      "step": 11990
    },
    {
      "epoch": 0.864340805881929,
      "grad_norm": 0.11796340326597786,
      "learning_rate": 9.499012153480925e-06,
      "loss": 1.0401,
      "step": 11991
    },
    {
      "epoch": 0.8644128883442659,
      "grad_norm": 0.12115760803392857,
      "learning_rate": 9.48908296267964e-06,
      "loss": 1.0741,
      "step": 11992
    },
    {
      "epoch": 0.8644849708066028,
      "grad_norm": 0.11513471292041728,
      "learning_rate": 9.479158705533075e-06,
      "loss": 1.0028,
      "step": 11993
    },
    {
      "epoch": 0.8645570532689396,
      "grad_norm": 0.11566446035020712,
      "learning_rate": 9.469239382582174e-06,
      "loss": 1.0408,
      "step": 11994
    },
    {
      "epoch": 0.8646291357312765,
      "grad_norm": 0.10980255264765325,
      "learning_rate": 9.459324994367657e-06,
      "loss": 0.9794,
      "step": 11995
    },
    {
      "epoch": 0.8647012181936135,
      "grad_norm": 0.1326941567448584,
      "learning_rate": 9.449415541429896e-06,
      "loss": 1.0428,
      "step": 11996
    },
    {
      "epoch": 0.8647733006559504,
      "grad_norm": 0.1277736140835412,
      "learning_rate": 9.439511024309088e-06,
      "loss": 1.0723,
      "step": 11997
    },
    {
      "epoch": 0.8648453831182873,
      "grad_norm": 0.12967556615768983,
      "learning_rate": 9.429611443545117e-06,
      "loss": 1.0828,
      "step": 11998
    },
    {
      "epoch": 0.8649174655806242,
      "grad_norm": 0.11314386686001082,
      "learning_rate": 9.419716799677603e-06,
      "loss": 0.9757,
      "step": 11999
    },
    {
      "epoch": 0.8649895480429611,
      "grad_norm": 0.13467612976088988,
      "learning_rate": 9.409827093245837e-06,
      "loss": 1.031,
      "step": 12000
    },
    {
      "epoch": 0.865061630505298,
      "grad_norm": 0.11282226589165448,
      "learning_rate": 9.399942324788957e-06,
      "loss": 0.9666,
      "step": 12001
    },
    {
      "epoch": 0.865133712967635,
      "grad_norm": 0.12465495141568667,
      "learning_rate": 9.390062494845753e-06,
      "loss": 1.0328,
      "step": 12002
    },
    {
      "epoch": 0.8652057954299719,
      "grad_norm": 0.11174307234225163,
      "learning_rate": 9.380187603954771e-06,
      "loss": 1.0103,
      "step": 12003
    },
    {
      "epoch": 0.8652778778923088,
      "grad_norm": 0.16362197811752235,
      "learning_rate": 9.370317652654248e-06,
      "loss": 1.0868,
      "step": 12004
    },
    {
      "epoch": 0.8653499603546457,
      "grad_norm": 0.18008780862233975,
      "learning_rate": 9.36045264148222e-06,
      "loss": 1.0671,
      "step": 12005
    },
    {
      "epoch": 0.8654220428169827,
      "grad_norm": 0.11157907903940029,
      "learning_rate": 9.350592570976414e-06,
      "loss": 1.0164,
      "step": 12006
    },
    {
      "epoch": 0.8654941252793196,
      "grad_norm": 0.17468566396110732,
      "learning_rate": 9.340737441674307e-06,
      "loss": 1.0375,
      "step": 12007
    },
    {
      "epoch": 0.8655662077416565,
      "grad_norm": 0.11412845538948922,
      "learning_rate": 9.33088725411304e-06,
      "loss": 1.0153,
      "step": 12008
    },
    {
      "epoch": 0.8656382902039934,
      "grad_norm": 0.13071859154138757,
      "learning_rate": 9.321042008829562e-06,
      "loss": 1.0342,
      "step": 12009
    },
    {
      "epoch": 0.8657103726663303,
      "grad_norm": 0.11552966561422383,
      "learning_rate": 9.311201706360539e-06,
      "loss": 1.0175,
      "step": 12010
    },
    {
      "epoch": 0.8657824551286672,
      "grad_norm": 0.1334351529358145,
      "learning_rate": 9.301366347242369e-06,
      "loss": 1.0906,
      "step": 12011
    },
    {
      "epoch": 0.8658545375910041,
      "grad_norm": 0.1154358369583546,
      "learning_rate": 9.291535932011131e-06,
      "loss": 1.0391,
      "step": 12012
    },
    {
      "epoch": 0.865926620053341,
      "grad_norm": 0.11100743905776819,
      "learning_rate": 9.281710461202664e-06,
      "loss": 0.9751,
      "step": 12013
    },
    {
      "epoch": 0.8659987025156779,
      "grad_norm": 0.15450828273235187,
      "learning_rate": 9.271889935352595e-06,
      "loss": 1.0428,
      "step": 12014
    },
    {
      "epoch": 0.8660707849780148,
      "grad_norm": 0.12302927665753777,
      "learning_rate": 9.262074354996209e-06,
      "loss": 1.0033,
      "step": 12015
    },
    {
      "epoch": 0.8661428674403517,
      "grad_norm": 0.13579022179871714,
      "learning_rate": 9.25226372066852e-06,
      "loss": 1.0699,
      "step": 12016
    },
    {
      "epoch": 0.8662149499026887,
      "grad_norm": 0.13805341628686282,
      "learning_rate": 9.242458032904311e-06,
      "loss": 1.0171,
      "step": 12017
    },
    {
      "epoch": 0.8662870323650256,
      "grad_norm": 0.14131949204515565,
      "learning_rate": 9.232657292238055e-06,
      "loss": 0.9902,
      "step": 12018
    },
    {
      "epoch": 0.8663591148273625,
      "grad_norm": 0.11717387879309889,
      "learning_rate": 9.222861499204049e-06,
      "loss": 1.0652,
      "step": 12019
    },
    {
      "epoch": 0.8664311972896994,
      "grad_norm": 0.13865873249680433,
      "learning_rate": 9.213070654336187e-06,
      "loss": 1.0811,
      "step": 12020
    },
    {
      "epoch": 0.8665032797520363,
      "grad_norm": 0.11328089972012079,
      "learning_rate": 9.203284758168173e-06,
      "loss": 1.0174,
      "step": 12021
    },
    {
      "epoch": 0.8665753622143733,
      "grad_norm": 0.12051750169774209,
      "learning_rate": 9.193503811233417e-06,
      "loss": 1.0378,
      "step": 12022
    },
    {
      "epoch": 0.8666474446767102,
      "grad_norm": 0.1364793857765151,
      "learning_rate": 9.183727814065112e-06,
      "loss": 0.9843,
      "step": 12023
    },
    {
      "epoch": 0.8667195271390471,
      "grad_norm": 0.12344940160266421,
      "learning_rate": 9.17395676719609e-06,
      "loss": 0.9774,
      "step": 12024
    },
    {
      "epoch": 0.866791609601384,
      "grad_norm": 0.4099984684551303,
      "learning_rate": 9.164190671158979e-06,
      "loss": 1.0313,
      "step": 12025
    },
    {
      "epoch": 0.8668636920637209,
      "grad_norm": 0.12479981062205693,
      "learning_rate": 9.154429526486098e-06,
      "loss": 1.0556,
      "step": 12026
    },
    {
      "epoch": 0.8669357745260579,
      "grad_norm": 0.1331971392453474,
      "learning_rate": 9.144673333709575e-06,
      "loss": 1.083,
      "step": 12027
    },
    {
      "epoch": 0.8670078569883948,
      "grad_norm": 0.11863643606198619,
      "learning_rate": 9.134922093361154e-06,
      "loss": 1.0205,
      "step": 12028
    },
    {
      "epoch": 0.8670799394507316,
      "grad_norm": 0.16306485985054564,
      "learning_rate": 9.125175805972375e-06,
      "loss": 1.0216,
      "step": 12029
    },
    {
      "epoch": 0.8671520219130685,
      "grad_norm": 0.12190842472674344,
      "learning_rate": 9.115434472074503e-06,
      "loss": 1.0371,
      "step": 12030
    },
    {
      "epoch": 0.8672241043754054,
      "grad_norm": 0.1176864981755896,
      "learning_rate": 9.105698092198533e-06,
      "loss": 1.0278,
      "step": 12031
    },
    {
      "epoch": 0.8672961868377423,
      "grad_norm": 0.12037700131855121,
      "learning_rate": 9.095966666875178e-06,
      "loss": 1.0819,
      "step": 12032
    },
    {
      "epoch": 0.8673682693000793,
      "grad_norm": 0.19704018553396785,
      "learning_rate": 9.086240196634898e-06,
      "loss": 1.0159,
      "step": 12033
    },
    {
      "epoch": 0.8674403517624162,
      "grad_norm": 0.340621292683631,
      "learning_rate": 9.07651868200785e-06,
      "loss": 1.1098,
      "step": 12034
    },
    {
      "epoch": 0.8675124342247531,
      "grad_norm": 0.1302900341784528,
      "learning_rate": 9.066802123523977e-06,
      "loss": 1.017,
      "step": 12035
    },
    {
      "epoch": 0.86758451668709,
      "grad_norm": 0.12345981583708511,
      "learning_rate": 9.057090521712886e-06,
      "loss": 0.9348,
      "step": 12036
    },
    {
      "epoch": 0.8676565991494269,
      "grad_norm": 0.12361348438816383,
      "learning_rate": 9.047383877103965e-06,
      "loss": 1.0019,
      "step": 12037
    },
    {
      "epoch": 0.8677286816117639,
      "grad_norm": 0.11842178566120608,
      "learning_rate": 9.037682190226304e-06,
      "loss": 0.8842,
      "step": 12038
    },
    {
      "epoch": 0.8678007640741008,
      "grad_norm": 0.11889135587797489,
      "learning_rate": 9.027985461608746e-06,
      "loss": 1.0163,
      "step": 12039
    },
    {
      "epoch": 0.8678728465364377,
      "grad_norm": 0.12452052317506371,
      "learning_rate": 9.018293691779833e-06,
      "loss": 1.0366,
      "step": 12040
    },
    {
      "epoch": 0.8679449289987746,
      "grad_norm": 0.52052204991447,
      "learning_rate": 9.008606881267856e-06,
      "loss": 1.0195,
      "step": 12041
    },
    {
      "epoch": 0.8680170114611115,
      "grad_norm": 0.12130320410416133,
      "learning_rate": 8.998925030600835e-06,
      "loss": 1.0597,
      "step": 12042
    },
    {
      "epoch": 0.8680890939234485,
      "grad_norm": 0.13346216063146896,
      "learning_rate": 8.989248140306527e-06,
      "loss": 1.0771,
      "step": 12043
    },
    {
      "epoch": 0.8681611763857854,
      "grad_norm": 0.11264216081487229,
      "learning_rate": 8.979576210912399e-06,
      "loss": 0.9987,
      "step": 12044
    },
    {
      "epoch": 0.8682332588481223,
      "grad_norm": 0.11238133981299013,
      "learning_rate": 8.969909242945651e-06,
      "loss": 0.9912,
      "step": 12045
    },
    {
      "epoch": 0.8683053413104591,
      "grad_norm": 0.10837985253712962,
      "learning_rate": 8.96024723693324e-06,
      "loss": 1.0625,
      "step": 12046
    },
    {
      "epoch": 0.868377423772796,
      "grad_norm": 0.12911982832674046,
      "learning_rate": 8.95059019340181e-06,
      "loss": 1.037,
      "step": 12047
    },
    {
      "epoch": 0.868449506235133,
      "grad_norm": 0.12440800490186052,
      "learning_rate": 8.940938112877772e-06,
      "loss": 1.0705,
      "step": 12048
    },
    {
      "epoch": 0.8685215886974699,
      "grad_norm": 0.12577874338671474,
      "learning_rate": 8.931290995887242e-06,
      "loss": 1.0105,
      "step": 12049
    },
    {
      "epoch": 0.8685936711598068,
      "grad_norm": 0.12605963614006743,
      "learning_rate": 8.921648842956077e-06,
      "loss": 1.018,
      "step": 12050
    },
    {
      "epoch": 0.8686657536221437,
      "grad_norm": 0.1129515517140717,
      "learning_rate": 8.912011654609852e-06,
      "loss": 0.9786,
      "step": 12051
    },
    {
      "epoch": 0.8687378360844806,
      "grad_norm": 0.12423152249505838,
      "learning_rate": 8.90237943137392e-06,
      "loss": 1.0296,
      "step": 12052
    },
    {
      "epoch": 0.8688099185468176,
      "grad_norm": 0.13009782291713737,
      "learning_rate": 8.892752173773245e-06,
      "loss": 1.0722,
      "step": 12053
    },
    {
      "epoch": 0.8688820010091545,
      "grad_norm": 0.15670284122203323,
      "learning_rate": 8.883129882332675e-06,
      "loss": 0.9924,
      "step": 12054
    },
    {
      "epoch": 0.8689540834714914,
      "grad_norm": 0.11045236612123095,
      "learning_rate": 8.873512557576668e-06,
      "loss": 1.0413,
      "step": 12055
    },
    {
      "epoch": 0.8690261659338283,
      "grad_norm": 0.1249013931825028,
      "learning_rate": 8.863900200029495e-06,
      "loss": 1.0189,
      "step": 12056
    },
    {
      "epoch": 0.8690982483961652,
      "grad_norm": 0.2576819222768655,
      "learning_rate": 8.854292810215048e-06,
      "loss": 1.0037,
      "step": 12057
    },
    {
      "epoch": 0.8691703308585021,
      "grad_norm": 0.12436110455455328,
      "learning_rate": 8.844690388657073e-06,
      "loss": 0.9726,
      "step": 12058
    },
    {
      "epoch": 0.8692424133208391,
      "grad_norm": 0.13971024544498634,
      "learning_rate": 8.835092935878974e-06,
      "loss": 1.0131,
      "step": 12059
    },
    {
      "epoch": 0.869314495783176,
      "grad_norm": 0.1499818909308736,
      "learning_rate": 8.825500452403901e-06,
      "loss": 1.0419,
      "step": 12060
    },
    {
      "epoch": 0.8693865782455129,
      "grad_norm": 0.1252382421765213,
      "learning_rate": 8.815912938754722e-06,
      "loss": 1.0764,
      "step": 12061
    },
    {
      "epoch": 0.8694586607078498,
      "grad_norm": 0.12115712274386467,
      "learning_rate": 8.80633039545402e-06,
      "loss": 1.0655,
      "step": 12062
    },
    {
      "epoch": 0.8695307431701867,
      "grad_norm": 0.2637551982541707,
      "learning_rate": 8.796752823024168e-06,
      "loss": 1.1137,
      "step": 12063
    },
    {
      "epoch": 0.8696028256325236,
      "grad_norm": 0.15220105718582494,
      "learning_rate": 8.787180221987235e-06,
      "loss": 1.0566,
      "step": 12064
    },
    {
      "epoch": 0.8696749080948605,
      "grad_norm": 0.13030122754196866,
      "learning_rate": 8.77761259286497e-06,
      "loss": 1.1169,
      "step": 12065
    },
    {
      "epoch": 0.8697469905571974,
      "grad_norm": 0.13433479814925164,
      "learning_rate": 8.768049936178913e-06,
      "loss": 1.0499,
      "step": 12066
    },
    {
      "epoch": 0.8698190730195343,
      "grad_norm": 0.129305614510485,
      "learning_rate": 8.758492252450324e-06,
      "loss": 1.0403,
      "step": 12067
    },
    {
      "epoch": 0.8698911554818712,
      "grad_norm": 0.11559718675944147,
      "learning_rate": 8.748939542200196e-06,
      "loss": 1.0643,
      "step": 12068
    },
    {
      "epoch": 0.8699632379442082,
      "grad_norm": 0.17872583353345842,
      "learning_rate": 8.739391805949205e-06,
      "loss": 1.0647,
      "step": 12069
    },
    {
      "epoch": 0.8700353204065451,
      "grad_norm": 0.13986178960768184,
      "learning_rate": 8.729849044217786e-06,
      "loss": 1.0475,
      "step": 12070
    },
    {
      "epoch": 0.870107402868882,
      "grad_norm": 0.12419741422898951,
      "learning_rate": 8.720311257526126e-06,
      "loss": 1.03,
      "step": 12071
    },
    {
      "epoch": 0.8701794853312189,
      "grad_norm": 0.1338914866087035,
      "learning_rate": 8.710778446394142e-06,
      "loss": 1.1593,
      "step": 12072
    },
    {
      "epoch": 0.8702515677935558,
      "grad_norm": 0.12049568505349868,
      "learning_rate": 8.701250611341416e-06,
      "loss": 0.9884,
      "step": 12073
    },
    {
      "epoch": 0.8703236502558928,
      "grad_norm": 0.130683090654487,
      "learning_rate": 8.691727752887303e-06,
      "loss": 1.0497,
      "step": 12074
    },
    {
      "epoch": 0.8703957327182297,
      "grad_norm": 0.4637195825447682,
      "learning_rate": 8.682209871550884e-06,
      "loss": 1.0818,
      "step": 12075
    },
    {
      "epoch": 0.8704678151805666,
      "grad_norm": 0.11062081260987021,
      "learning_rate": 8.672696967851013e-06,
      "loss": 1.0312,
      "step": 12076
    },
    {
      "epoch": 0.8705398976429035,
      "grad_norm": 0.12621012288945288,
      "learning_rate": 8.663189042306174e-06,
      "loss": 1.0001,
      "step": 12077
    },
    {
      "epoch": 0.8706119801052404,
      "grad_norm": 0.11584454943472587,
      "learning_rate": 8.653686095434655e-06,
      "loss": 1.0695,
      "step": 12078
    },
    {
      "epoch": 0.8706840625675774,
      "grad_norm": 0.1351576430245374,
      "learning_rate": 8.644188127754427e-06,
      "loss": 1.0898,
      "step": 12079
    },
    {
      "epoch": 0.8707561450299143,
      "grad_norm": 0.12315550213583978,
      "learning_rate": 8.63469513978329e-06,
      "loss": 1.0461,
      "step": 12080
    },
    {
      "epoch": 0.8708282274922511,
      "grad_norm": 0.11840934691078098,
      "learning_rate": 8.625207132038615e-06,
      "loss": 1.0849,
      "step": 12081
    },
    {
      "epoch": 0.870900309954588,
      "grad_norm": 0.13166532149617047,
      "learning_rate": 8.615724105037615e-06,
      "loss": 1.0484,
      "step": 12082
    },
    {
      "epoch": 0.8709723924169249,
      "grad_norm": 0.15381925592312062,
      "learning_rate": 8.606246059297207e-06,
      "loss": 1.071,
      "step": 12083
    },
    {
      "epoch": 0.8710444748792618,
      "grad_norm": 0.11950402663901266,
      "learning_rate": 8.596772995334012e-06,
      "loss": 1.0528,
      "step": 12084
    },
    {
      "epoch": 0.8711165573415988,
      "grad_norm": 0.12177216540421525,
      "learning_rate": 8.587304913664407e-06,
      "loss": 1.0811,
      "step": 12085
    },
    {
      "epoch": 0.8711886398039357,
      "grad_norm": 0.11023628767617381,
      "learning_rate": 8.577841814804489e-06,
      "loss": 1.0062,
      "step": 12086
    },
    {
      "epoch": 0.8712607222662726,
      "grad_norm": 0.13052589688031033,
      "learning_rate": 8.568383699270077e-06,
      "loss": 1.1144,
      "step": 12087
    },
    {
      "epoch": 0.8713328047286095,
      "grad_norm": 0.14519301325624742,
      "learning_rate": 8.558930567576718e-06,
      "loss": 0.9987,
      "step": 12088
    },
    {
      "epoch": 0.8714048871909464,
      "grad_norm": 0.12099975053598458,
      "learning_rate": 8.549482420239707e-06,
      "loss": 1.034,
      "step": 12089
    },
    {
      "epoch": 0.8714769696532834,
      "grad_norm": 0.12441885294526535,
      "learning_rate": 8.540039257774035e-06,
      "loss": 1.0491,
      "step": 12090
    },
    {
      "epoch": 0.8715490521156203,
      "grad_norm": 0.12185803628170468,
      "learning_rate": 8.530601080694456e-06,
      "loss": 1.0542,
      "step": 12091
    },
    {
      "epoch": 0.8716211345779572,
      "grad_norm": 0.1482809749152308,
      "learning_rate": 8.521167889515424e-06,
      "loss": 1.0393,
      "step": 12092
    },
    {
      "epoch": 0.8716932170402941,
      "grad_norm": 0.11415158811145261,
      "learning_rate": 8.51173968475113e-06,
      "loss": 1.0187,
      "step": 12093
    },
    {
      "epoch": 0.871765299502631,
      "grad_norm": 0.13767614032691766,
      "learning_rate": 8.502316466915505e-06,
      "loss": 1.1026,
      "step": 12094
    },
    {
      "epoch": 0.871837381964968,
      "grad_norm": 0.11494400777583037,
      "learning_rate": 8.492898236522196e-06,
      "loss": 0.9672,
      "step": 12095
    },
    {
      "epoch": 0.8719094644273049,
      "grad_norm": 0.12326473037066335,
      "learning_rate": 8.483484994084578e-06,
      "loss": 0.9412,
      "step": 12096
    },
    {
      "epoch": 0.8719815468896418,
      "grad_norm": 0.12482392667124868,
      "learning_rate": 8.474076740115754e-06,
      "loss": 0.9714,
      "step": 12097
    },
    {
      "epoch": 0.8720536293519786,
      "grad_norm": 0.11037467357333978,
      "learning_rate": 8.464673475128571e-06,
      "loss": 1.0132,
      "step": 12098
    },
    {
      "epoch": 0.8721257118143155,
      "grad_norm": 0.11880339372953963,
      "learning_rate": 8.455275199635571e-06,
      "loss": 1.0151,
      "step": 12099
    },
    {
      "epoch": 0.8721977942766524,
      "grad_norm": 0.11531450675750775,
      "learning_rate": 8.445881914149068e-06,
      "loss": 1.0481,
      "step": 12100
    },
    {
      "epoch": 0.8722698767389894,
      "grad_norm": 0.11780873834169278,
      "learning_rate": 8.436493619181051e-06,
      "loss": 1.0748,
      "step": 12101
    },
    {
      "epoch": 0.8723419592013263,
      "grad_norm": 0.11242145787357026,
      "learning_rate": 8.427110315243291e-06,
      "loss": 0.9463,
      "step": 12102
    },
    {
      "epoch": 0.8724140416636632,
      "grad_norm": 0.1631629722066391,
      "learning_rate": 8.417732002847255e-06,
      "loss": 1.1216,
      "step": 12103
    },
    {
      "epoch": 0.8724861241260001,
      "grad_norm": 0.13606348099003285,
      "learning_rate": 8.408358682504147e-06,
      "loss": 1.0441,
      "step": 12104
    },
    {
      "epoch": 0.872558206588337,
      "grad_norm": 0.12671485050033204,
      "learning_rate": 8.398990354724901e-06,
      "loss": 0.9558,
      "step": 12105
    },
    {
      "epoch": 0.872630289050674,
      "grad_norm": 0.16124689053716837,
      "learning_rate": 8.38962702002013e-06,
      "loss": 1.0679,
      "step": 12106
    },
    {
      "epoch": 0.8727023715130109,
      "grad_norm": 0.11641403792448997,
      "learning_rate": 8.380268678900283e-06,
      "loss": 1.0905,
      "step": 12107
    },
    {
      "epoch": 0.8727744539753478,
      "grad_norm": 0.11087083902562947,
      "learning_rate": 8.37091533187544e-06,
      "loss": 0.995,
      "step": 12108
    },
    {
      "epoch": 0.8728465364376847,
      "grad_norm": 0.11673810721946587,
      "learning_rate": 8.361566979455471e-06,
      "loss": 1.0223,
      "step": 12109
    },
    {
      "epoch": 0.8729186189000216,
      "grad_norm": 0.12573684821592163,
      "learning_rate": 8.35222362214989e-06,
      "loss": 1.0144,
      "step": 12110
    },
    {
      "epoch": 0.8729907013623586,
      "grad_norm": 0.1306435159197036,
      "learning_rate": 8.342885260468037e-06,
      "loss": 0.9525,
      "step": 12111
    },
    {
      "epoch": 0.8730627838246955,
      "grad_norm": 0.3157283194780034,
      "learning_rate": 8.333551894918924e-06,
      "loss": 1.0704,
      "step": 12112
    },
    {
      "epoch": 0.8731348662870324,
      "grad_norm": 0.11963684112485698,
      "learning_rate": 8.324223526011321e-06,
      "loss": 1.041,
      "step": 12113
    },
    {
      "epoch": 0.8732069487493693,
      "grad_norm": 0.12874637694599053,
      "learning_rate": 8.314900154253669e-06,
      "loss": 1.0724,
      "step": 12114
    },
    {
      "epoch": 0.8732790312117062,
      "grad_norm": 0.4213041009084777,
      "learning_rate": 8.305581780154204e-06,
      "loss": 1.0104,
      "step": 12115
    },
    {
      "epoch": 0.8733511136740431,
      "grad_norm": 0.1269352029592907,
      "learning_rate": 8.296268404220865e-06,
      "loss": 1.0409,
      "step": 12116
    },
    {
      "epoch": 0.87342319613638,
      "grad_norm": 0.13508607829901786,
      "learning_rate": 8.286960026961299e-06,
      "loss": 1.0364,
      "step": 12117
    },
    {
      "epoch": 0.8734952785987169,
      "grad_norm": 0.1265535486702728,
      "learning_rate": 8.277656648882893e-06,
      "loss": 0.998,
      "step": 12118
    },
    {
      "epoch": 0.8735673610610538,
      "grad_norm": 0.11598598571826521,
      "learning_rate": 8.268358270492749e-06,
      "loss": 0.9525,
      "step": 12119
    },
    {
      "epoch": 0.8736394435233907,
      "grad_norm": 0.11450730167759086,
      "learning_rate": 8.259064892297775e-06,
      "loss": 0.9742,
      "step": 12120
    },
    {
      "epoch": 0.8737115259857277,
      "grad_norm": 0.10801883133371241,
      "learning_rate": 8.249776514804485e-06,
      "loss": 1.0508,
      "step": 12121
    },
    {
      "epoch": 0.8737836084480646,
      "grad_norm": 0.1280418053736943,
      "learning_rate": 8.2404931385192e-06,
      "loss": 1.0266,
      "step": 12122
    },
    {
      "epoch": 0.8738556909104015,
      "grad_norm": 0.13390854570171942,
      "learning_rate": 8.231214763947914e-06,
      "loss": 1.0533,
      "step": 12123
    },
    {
      "epoch": 0.8739277733727384,
      "grad_norm": 0.12972911661419143,
      "learning_rate": 8.221941391596454e-06,
      "loss": 1.0581,
      "step": 12124
    },
    {
      "epoch": 0.8739998558350753,
      "grad_norm": 0.11417092008232281,
      "learning_rate": 8.21267302197024e-06,
      "loss": 0.9781,
      "step": 12125
    },
    {
      "epoch": 0.8740719382974123,
      "grad_norm": 0.13334767979345258,
      "learning_rate": 8.20340965557449e-06,
      "loss": 1.0334,
      "step": 12126
    },
    {
      "epoch": 0.8741440207597492,
      "grad_norm": 0.21398954379294982,
      "learning_rate": 8.194151292914166e-06,
      "loss": 1.0365,
      "step": 12127
    },
    {
      "epoch": 0.8742161032220861,
      "grad_norm": 0.124177190093061,
      "learning_rate": 8.184897934493907e-06,
      "loss": 1.0076,
      "step": 12128
    },
    {
      "epoch": 0.874288185684423,
      "grad_norm": 0.15487534608113077,
      "learning_rate": 8.1756495808181e-06,
      "loss": 0.986,
      "step": 12129
    },
    {
      "epoch": 0.8743602681467599,
      "grad_norm": 1.0736175484041395,
      "learning_rate": 8.166406232390889e-06,
      "loss": 1.0219,
      "step": 12130
    },
    {
      "epoch": 0.8744323506090969,
      "grad_norm": 0.12073927766578513,
      "learning_rate": 8.157167889716112e-06,
      "loss": 1.0151,
      "step": 12131
    },
    {
      "epoch": 0.8745044330714338,
      "grad_norm": 0.16760646905611185,
      "learning_rate": 8.14793455329732e-06,
      "loss": 1.0854,
      "step": 12132
    },
    {
      "epoch": 0.8745765155337706,
      "grad_norm": 0.12632435457654437,
      "learning_rate": 8.138706223637827e-06,
      "loss": 1.0062,
      "step": 12133
    },
    {
      "epoch": 0.8746485979961075,
      "grad_norm": 0.1462842676058375,
      "learning_rate": 8.12948290124067e-06,
      "loss": 0.974,
      "step": 12134
    },
    {
      "epoch": 0.8747206804584444,
      "grad_norm": 0.15784504323991871,
      "learning_rate": 8.120264586608584e-06,
      "loss": 1.0069,
      "step": 12135
    },
    {
      "epoch": 0.8747927629207813,
      "grad_norm": 0.13128531917848235,
      "learning_rate": 8.111051280244053e-06,
      "loss": 1.0283,
      "step": 12136
    },
    {
      "epoch": 0.8748648453831183,
      "grad_norm": 0.12844596691130425,
      "learning_rate": 8.1018429826493e-06,
      "loss": 0.9885,
      "step": 12137
    },
    {
      "epoch": 0.8749369278454552,
      "grad_norm": 0.11655669299902191,
      "learning_rate": 8.092639694326242e-06,
      "loss": 1.1185,
      "step": 12138
    },
    {
      "epoch": 0.8750090103077921,
      "grad_norm": 0.11386858809730575,
      "learning_rate": 8.08344141577655e-06,
      "loss": 1.065,
      "step": 12139
    },
    {
      "epoch": 0.875081092770129,
      "grad_norm": 0.12832996755952658,
      "learning_rate": 8.074248147501617e-06,
      "loss": 1.0749,
      "step": 12140
    },
    {
      "epoch": 0.8751531752324659,
      "grad_norm": 0.11058911280412045,
      "learning_rate": 8.065059890002547e-06,
      "loss": 0.9802,
      "step": 12141
    },
    {
      "epoch": 0.8752252576948029,
      "grad_norm": 0.15843442165965763,
      "learning_rate": 8.05587664378018e-06,
      "loss": 1.041,
      "step": 12142
    },
    {
      "epoch": 0.8752973401571398,
      "grad_norm": 0.14725016119653955,
      "learning_rate": 8.046698409335096e-06,
      "loss": 1.0417,
      "step": 12143
    },
    {
      "epoch": 0.8753694226194767,
      "grad_norm": 0.1349156643843978,
      "learning_rate": 8.037525187167583e-06,
      "loss": 1.0549,
      "step": 12144
    },
    {
      "epoch": 0.8754415050818136,
      "grad_norm": 0.12488706445911088,
      "learning_rate": 8.028356977777674e-06,
      "loss": 1.0525,
      "step": 12145
    },
    {
      "epoch": 0.8755135875441505,
      "grad_norm": 0.12982026340245828,
      "learning_rate": 8.019193781665103e-06,
      "loss": 1.0801,
      "step": 12146
    },
    {
      "epoch": 0.8755856700064875,
      "grad_norm": 0.13195291070987464,
      "learning_rate": 8.010035599329358e-06,
      "loss": 0.9761,
      "step": 12147
    },
    {
      "epoch": 0.8756577524688244,
      "grad_norm": 0.12452120436512931,
      "learning_rate": 8.000882431269641e-06,
      "loss": 1.0001,
      "step": 12148
    },
    {
      "epoch": 0.8757298349311613,
      "grad_norm": 0.14742536949132234,
      "learning_rate": 7.991734277984897e-06,
      "loss": 1.0843,
      "step": 12149
    },
    {
      "epoch": 0.8758019173934981,
      "grad_norm": 0.2099364678930831,
      "learning_rate": 7.982591139973727e-06,
      "loss": 1.0466,
      "step": 12150
    },
    {
      "epoch": 0.875873999855835,
      "grad_norm": 0.2975873126161979,
      "learning_rate": 7.973453017734567e-06,
      "loss": 1.096,
      "step": 12151
    },
    {
      "epoch": 0.875946082318172,
      "grad_norm": 0.12022273567539925,
      "learning_rate": 7.964319911765517e-06,
      "loss": 1.0417,
      "step": 12152
    },
    {
      "epoch": 0.8760181647805089,
      "grad_norm": 0.11011058773392632,
      "learning_rate": 7.955191822564412e-06,
      "loss": 0.9602,
      "step": 12153
    },
    {
      "epoch": 0.8760902472428458,
      "grad_norm": 0.16681799773003195,
      "learning_rate": 7.946068750628788e-06,
      "loss": 1.0057,
      "step": 12154
    },
    {
      "epoch": 0.8761623297051827,
      "grad_norm": 0.12235132750266613,
      "learning_rate": 7.93695069645597e-06,
      "loss": 1.035,
      "step": 12155
    },
    {
      "epoch": 0.8762344121675196,
      "grad_norm": 0.1378652340756323,
      "learning_rate": 7.927837660542947e-06,
      "loss": 1.0907,
      "step": 12156
    },
    {
      "epoch": 0.8763064946298565,
      "grad_norm": 0.1279973259624165,
      "learning_rate": 7.918729643386503e-06,
      "loss": 1.0524,
      "step": 12157
    },
    {
      "epoch": 0.8763785770921935,
      "grad_norm": 0.16768558161668654,
      "learning_rate": 7.909626645483037e-06,
      "loss": 1.0647,
      "step": 12158
    },
    {
      "epoch": 0.8764506595545304,
      "grad_norm": 0.17213890085996378,
      "learning_rate": 7.900528667328799e-06,
      "loss": 1.0352,
      "step": 12159
    },
    {
      "epoch": 0.8765227420168673,
      "grad_norm": 0.13682907947634262,
      "learning_rate": 7.891435709419691e-06,
      "loss": 1.0741,
      "step": 12160
    },
    {
      "epoch": 0.8765948244792042,
      "grad_norm": 0.16892605727401253,
      "learning_rate": 7.882347772251386e-06,
      "loss": 1.0294,
      "step": 12161
    },
    {
      "epoch": 0.8766669069415411,
      "grad_norm": 0.11908994926029752,
      "learning_rate": 7.873264856319218e-06,
      "loss": 1.0523,
      "step": 12162
    },
    {
      "epoch": 0.8767389894038781,
      "grad_norm": 0.12686129275704497,
      "learning_rate": 7.86418696211828e-06,
      "loss": 1.043,
      "step": 12163
    },
    {
      "epoch": 0.876811071866215,
      "grad_norm": 0.5797931854589491,
      "learning_rate": 7.855114090143446e-06,
      "loss": 1.092,
      "step": 12164
    },
    {
      "epoch": 0.8768831543285519,
      "grad_norm": 0.3307779671248921,
      "learning_rate": 7.84604624088927e-06,
      "loss": 1.0134,
      "step": 12165
    },
    {
      "epoch": 0.8769552367908888,
      "grad_norm": 0.1098000954061004,
      "learning_rate": 7.836983414849975e-06,
      "loss": 0.9789,
      "step": 12166
    },
    {
      "epoch": 0.8770273192532257,
      "grad_norm": 0.12375881616315769,
      "learning_rate": 7.827925612519594e-06,
      "loss": 1.0152,
      "step": 12167
    },
    {
      "epoch": 0.8770994017155626,
      "grad_norm": 0.2213641801469447,
      "learning_rate": 7.818872834391877e-06,
      "loss": 1.0095,
      "step": 12168
    },
    {
      "epoch": 0.8771714841778995,
      "grad_norm": 0.11648706237480513,
      "learning_rate": 7.809825080960276e-06,
      "loss": 1.0947,
      "step": 12169
    },
    {
      "epoch": 0.8772435666402364,
      "grad_norm": 0.11347399611614506,
      "learning_rate": 7.80078235271795e-06,
      "loss": 1.0567,
      "step": 12170
    },
    {
      "epoch": 0.8773156491025733,
      "grad_norm": 0.11054590189065305,
      "learning_rate": 7.791744650157829e-06,
      "loss": 1.0319,
      "step": 12171
    },
    {
      "epoch": 0.8773877315649102,
      "grad_norm": 0.11814368572893381,
      "learning_rate": 7.782711973772516e-06,
      "loss": 1.0267,
      "step": 12172
    },
    {
      "epoch": 0.8774598140272472,
      "grad_norm": 0.20122044514239304,
      "learning_rate": 7.773684324054442e-06,
      "loss": 1.1107,
      "step": 12173
    },
    {
      "epoch": 0.8775318964895841,
      "grad_norm": 0.14671069910092976,
      "learning_rate": 7.764661701495635e-06,
      "loss": 0.9835,
      "step": 12174
    },
    {
      "epoch": 0.877603978951921,
      "grad_norm": 1.2133963512963484,
      "learning_rate": 7.755644106587934e-06,
      "loss": 1.0761,
      "step": 12175
    },
    {
      "epoch": 0.8776760614142579,
      "grad_norm": 0.11057742685809638,
      "learning_rate": 7.746631539822835e-06,
      "loss": 1.0294,
      "step": 12176
    },
    {
      "epoch": 0.8777481438765948,
      "grad_norm": 0.2589896354647269,
      "learning_rate": 7.73762400169169e-06,
      "loss": 1.0594,
      "step": 12177
    },
    {
      "epoch": 0.8778202263389318,
      "grad_norm": 0.10342488709290061,
      "learning_rate": 7.728621492685418e-06,
      "loss": 1.0074,
      "step": 12178
    },
    {
      "epoch": 0.8778923088012687,
      "grad_norm": 0.1246004963933191,
      "learning_rate": 7.719624013294757e-06,
      "loss": 1.0358,
      "step": 12179
    },
    {
      "epoch": 0.8779643912636056,
      "grad_norm": 0.12306296824608992,
      "learning_rate": 7.710631564010129e-06,
      "loss": 1.0535,
      "step": 12180
    },
    {
      "epoch": 0.8780364737259425,
      "grad_norm": 0.11707397296766658,
      "learning_rate": 7.701644145321763e-06,
      "loss": 1.0419,
      "step": 12181
    },
    {
      "epoch": 0.8781085561882794,
      "grad_norm": 0.153130692859966,
      "learning_rate": 7.6926617577195e-06,
      "loss": 1.0131,
      "step": 12182
    },
    {
      "epoch": 0.8781806386506164,
      "grad_norm": 0.14431763449562449,
      "learning_rate": 7.68368440169297e-06,
      "loss": 1.054,
      "step": 12183
    },
    {
      "epoch": 0.8782527211129533,
      "grad_norm": 0.1947058432714164,
      "learning_rate": 7.674712077731538e-06,
      "loss": 1.0068,
      "step": 12184
    },
    {
      "epoch": 0.8783248035752901,
      "grad_norm": 0.18808754960145926,
      "learning_rate": 7.665744786324258e-06,
      "loss": 1.0464,
      "step": 12185
    },
    {
      "epoch": 0.878396886037627,
      "grad_norm": 0.11802246351690969,
      "learning_rate": 7.656782527959927e-06,
      "loss": 1.0794,
      "step": 12186
    },
    {
      "epoch": 0.8784689684999639,
      "grad_norm": 0.12337061579665337,
      "learning_rate": 7.647825303127076e-06,
      "loss": 1.0473,
      "step": 12187
    },
    {
      "epoch": 0.8785410509623008,
      "grad_norm": 0.11164736752006942,
      "learning_rate": 7.63887311231396e-06,
      "loss": 0.9936,
      "step": 12188
    },
    {
      "epoch": 0.8786131334246378,
      "grad_norm": 0.13083641866170515,
      "learning_rate": 7.629925956008533e-06,
      "loss": 1.0621,
      "step": 12189
    },
    {
      "epoch": 0.8786852158869747,
      "grad_norm": 0.11510603026784258,
      "learning_rate": 7.620983834698514e-06,
      "loss": 1.0167,
      "step": 12190
    },
    {
      "epoch": 0.8787572983493116,
      "grad_norm": 0.13517759754346745,
      "learning_rate": 7.612046748871327e-06,
      "loss": 1.0567,
      "step": 12191
    },
    {
      "epoch": 0.8788293808116485,
      "grad_norm": 0.1414864031978316,
      "learning_rate": 7.6031146990141134e-06,
      "loss": 1.0344,
      "step": 12192
    },
    {
      "epoch": 0.8789014632739854,
      "grad_norm": 0.10474736736780832,
      "learning_rate": 7.594187685613763e-06,
      "loss": 1.0327,
      "step": 12193
    },
    {
      "epoch": 0.8789735457363224,
      "grad_norm": 0.11756979165719424,
      "learning_rate": 7.585265709156864e-06,
      "loss": 1.0708,
      "step": 12194
    },
    {
      "epoch": 0.8790456281986593,
      "grad_norm": 0.18518061163264507,
      "learning_rate": 7.576348770129749e-06,
      "loss": 0.9647,
      "step": 12195
    },
    {
      "epoch": 0.8791177106609962,
      "grad_norm": 0.25565959176212427,
      "learning_rate": 7.567436869018462e-06,
      "loss": 1.0497,
      "step": 12196
    },
    {
      "epoch": 0.8791897931233331,
      "grad_norm": 0.3089001812619251,
      "learning_rate": 7.558530006308806e-06,
      "loss": 0.9641,
      "step": 12197
    },
    {
      "epoch": 0.87926187558567,
      "grad_norm": 0.11550596138584372,
      "learning_rate": 7.549628182486257e-06,
      "loss": 1.0623,
      "step": 12198
    },
    {
      "epoch": 0.879333958048007,
      "grad_norm": 0.12109090825099096,
      "learning_rate": 7.540731398036072e-06,
      "loss": 1.0406,
      "step": 12199
    },
    {
      "epoch": 0.8794060405103439,
      "grad_norm": 0.11505109948696854,
      "learning_rate": 7.531839653443174e-06,
      "loss": 1.0551,
      "step": 12200
    },
    {
      "epoch": 0.8794781229726808,
      "grad_norm": 0.14928482506513793,
      "learning_rate": 7.522952949192275e-06,
      "loss": 0.9981,
      "step": 12201
    },
    {
      "epoch": 0.8795502054350176,
      "grad_norm": 0.10586110304015121,
      "learning_rate": 7.514071285767766e-06,
      "loss": 0.969,
      "step": 12202
    },
    {
      "epoch": 0.8796222878973545,
      "grad_norm": 0.09562989279672607,
      "learning_rate": 7.505194663653747e-06,
      "loss": 0.9642,
      "step": 12203
    },
    {
      "epoch": 0.8796943703596914,
      "grad_norm": 0.12555993972747986,
      "learning_rate": 7.496323083334123e-06,
      "loss": 1.0212,
      "step": 12204
    },
    {
      "epoch": 0.8797664528220284,
      "grad_norm": 0.12131028825424232,
      "learning_rate": 7.487456545292437e-06,
      "loss": 1.065,
      "step": 12205
    },
    {
      "epoch": 0.8798385352843653,
      "grad_norm": 0.12918150131833311,
      "learning_rate": 7.478595050012038e-06,
      "loss": 0.9492,
      "step": 12206
    },
    {
      "epoch": 0.8799106177467022,
      "grad_norm": 0.1216237925325879,
      "learning_rate": 7.4697385979758946e-06,
      "loss": 1.0805,
      "step": 12207
    },
    {
      "epoch": 0.8799827002090391,
      "grad_norm": 0.1077740288937014,
      "learning_rate": 7.460887189666821e-06,
      "loss": 1.0301,
      "step": 12208
    },
    {
      "epoch": 0.880054782671376,
      "grad_norm": 0.11743549932727687,
      "learning_rate": 7.452040825567263e-06,
      "loss": 1.0314,
      "step": 12209
    },
    {
      "epoch": 0.880126865133713,
      "grad_norm": 0.12407976797775218,
      "learning_rate": 7.44319950615946e-06,
      "loss": 1.1337,
      "step": 12210
    },
    {
      "epoch": 0.8801989475960499,
      "grad_norm": 0.12418684067460713,
      "learning_rate": 7.434363231925279e-06,
      "loss": 1.0179,
      "step": 12211
    },
    {
      "epoch": 0.8802710300583868,
      "grad_norm": 0.12019213213270107,
      "learning_rate": 7.425532003346447e-06,
      "loss": 1.0838,
      "step": 12212
    },
    {
      "epoch": 0.8803431125207237,
      "grad_norm": 0.13692302435703096,
      "learning_rate": 7.416705820904313e-06,
      "loss": 1.0474,
      "step": 12213
    },
    {
      "epoch": 0.8804151949830606,
      "grad_norm": 0.11915206241868978,
      "learning_rate": 7.407884685079991e-06,
      "loss": 1.0774,
      "step": 12214
    },
    {
      "epoch": 0.8804872774453976,
      "grad_norm": 0.11559554465780338,
      "learning_rate": 7.399068596354297e-06,
      "loss": 1.0412,
      "step": 12215
    },
    {
      "epoch": 0.8805593599077345,
      "grad_norm": 0.13184842969048874,
      "learning_rate": 7.39025755520778e-06,
      "loss": 1.0146,
      "step": 12216
    },
    {
      "epoch": 0.8806314423700714,
      "grad_norm": 0.11782249248676337,
      "learning_rate": 7.381451562120756e-06,
      "loss": 0.973,
      "step": 12217
    },
    {
      "epoch": 0.8807035248324083,
      "grad_norm": 0.13364565556598318,
      "learning_rate": 7.372650617573229e-06,
      "loss": 1.0837,
      "step": 12218
    },
    {
      "epoch": 0.8807756072947451,
      "grad_norm": 0.1744677696173252,
      "learning_rate": 7.363854722044883e-06,
      "loss": 1.0605,
      "step": 12219
    },
    {
      "epoch": 0.880847689757082,
      "grad_norm": 0.111120223112922,
      "learning_rate": 7.3550638760152e-06,
      "loss": 0.9696,
      "step": 12220
    },
    {
      "epoch": 0.880919772219419,
      "grad_norm": 0.12119714363710639,
      "learning_rate": 7.3462780799633645e-06,
      "loss": 1.0219,
      "step": 12221
    },
    {
      "epoch": 0.8809918546817559,
      "grad_norm": 0.14718673780338354,
      "learning_rate": 7.337497334368304e-06,
      "loss": 1.0864,
      "step": 12222
    },
    {
      "epoch": 0.8810639371440928,
      "grad_norm": 0.11213712573955917,
      "learning_rate": 7.328721639708602e-06,
      "loss": 1.041,
      "step": 12223
    },
    {
      "epoch": 0.8811360196064297,
      "grad_norm": 0.11047501831258037,
      "learning_rate": 7.319950996462621e-06,
      "loss": 0.9404,
      "step": 12224
    },
    {
      "epoch": 0.8812081020687667,
      "grad_norm": 0.1107113222909177,
      "learning_rate": 7.311185405108445e-06,
      "loss": 1.0328,
      "step": 12225
    },
    {
      "epoch": 0.8812801845311036,
      "grad_norm": 0.11650237579683421,
      "learning_rate": 7.302424866123913e-06,
      "loss": 0.9804,
      "step": 12226
    },
    {
      "epoch": 0.8813522669934405,
      "grad_norm": 0.10920173382594489,
      "learning_rate": 7.2936693799865004e-06,
      "loss": 0.9873,
      "step": 12227
    },
    {
      "epoch": 0.8814243494557774,
      "grad_norm": 0.11541723753110279,
      "learning_rate": 7.2849189471734915e-06,
      "loss": 1.0939,
      "step": 12228
    },
    {
      "epoch": 0.8814964319181143,
      "grad_norm": 0.1281059026351098,
      "learning_rate": 7.276173568161826e-06,
      "loss": 1.0245,
      "step": 12229
    },
    {
      "epoch": 0.8815685143804513,
      "grad_norm": 0.12080264115157575,
      "learning_rate": 7.267433243428279e-06,
      "loss": 1.1003,
      "step": 12230
    },
    {
      "epoch": 0.8816405968427882,
      "grad_norm": 0.128580907300368,
      "learning_rate": 7.2586979734492025e-06,
      "loss": 1.0191,
      "step": 12231
    },
    {
      "epoch": 0.8817126793051251,
      "grad_norm": 0.12345806637432645,
      "learning_rate": 7.249967758700771e-06,
      "loss": 1.0768,
      "step": 12232
    },
    {
      "epoch": 0.881784761767462,
      "grad_norm": 0.130415928214622,
      "learning_rate": 7.241242599658849e-06,
      "loss": 0.9838,
      "step": 12233
    },
    {
      "epoch": 0.8818568442297989,
      "grad_norm": 0.14187335043715857,
      "learning_rate": 7.2325224967990876e-06,
      "loss": 1.0839,
      "step": 12234
    },
    {
      "epoch": 0.8819289266921359,
      "grad_norm": 0.1300510852892982,
      "learning_rate": 7.223807450596742e-06,
      "loss": 0.9918,
      "step": 12235
    },
    {
      "epoch": 0.8820010091544728,
      "grad_norm": 0.11744073609931133,
      "learning_rate": 7.215097461526898e-06,
      "loss": 1.0487,
      "step": 12236
    },
    {
      "epoch": 0.8820730916168096,
      "grad_norm": 0.11568903670682001,
      "learning_rate": 7.206392530064321e-06,
      "loss": 0.9554,
      "step": 12237
    },
    {
      "epoch": 0.8821451740791465,
      "grad_norm": 0.23348162473131936,
      "learning_rate": 7.1976926566835075e-06,
      "loss": 1.0647,
      "step": 12238
    },
    {
      "epoch": 0.8822172565414834,
      "grad_norm": 0.12300133126451648,
      "learning_rate": 7.18899784185868e-06,
      "loss": 1.0761,
      "step": 12239
    },
    {
      "epoch": 0.8822893390038203,
      "grad_norm": 0.12088571411925954,
      "learning_rate": 7.180308086063769e-06,
      "loss": 1.0621,
      "step": 12240
    },
    {
      "epoch": 0.8823614214661573,
      "grad_norm": 0.15500448946085693,
      "learning_rate": 7.171623389772464e-06,
      "loss": 1.0134,
      "step": 12241
    },
    {
      "epoch": 0.8824335039284942,
      "grad_norm": 0.13068108585740115,
      "learning_rate": 7.162943753458151e-06,
      "loss": 1.1332,
      "step": 12242
    },
    {
      "epoch": 0.8825055863908311,
      "grad_norm": 0.13064539213885876,
      "learning_rate": 7.154269177593953e-06,
      "loss": 0.988,
      "step": 12243
    },
    {
      "epoch": 0.882577668853168,
      "grad_norm": 0.15638195948246147,
      "learning_rate": 7.145599662652691e-06,
      "loss": 0.9986,
      "step": 12244
    },
    {
      "epoch": 0.8826497513155049,
      "grad_norm": 0.11034912084821337,
      "learning_rate": 7.136935209106954e-06,
      "loss": 1.0361,
      "step": 12245
    },
    {
      "epoch": 0.8827218337778419,
      "grad_norm": 0.11404265788713407,
      "learning_rate": 7.128275817429031e-06,
      "loss": 1.0119,
      "step": 12246
    },
    {
      "epoch": 0.8827939162401788,
      "grad_norm": 0.17401557292960937,
      "learning_rate": 7.119621488090922e-06,
      "loss": 1.0707,
      "step": 12247
    },
    {
      "epoch": 0.8828659987025157,
      "grad_norm": 0.12105871315298578,
      "learning_rate": 7.110972221564382e-06,
      "loss": 1.0047,
      "step": 12248
    },
    {
      "epoch": 0.8829380811648526,
      "grad_norm": 0.14429865475126358,
      "learning_rate": 7.102328018320858e-06,
      "loss": 1.0822,
      "step": 12249
    },
    {
      "epoch": 0.8830101636271895,
      "grad_norm": 0.13291627952743004,
      "learning_rate": 7.09368887883155e-06,
      "loss": 1.0438,
      "step": 12250
    },
    {
      "epoch": 0.8830822460895265,
      "grad_norm": 0.12027957595306386,
      "learning_rate": 7.085054803567348e-06,
      "loss": 1.068,
      "step": 12251
    },
    {
      "epoch": 0.8831543285518634,
      "grad_norm": 0.11841305827522278,
      "learning_rate": 7.076425792998909e-06,
      "loss": 1.0109,
      "step": 12252
    },
    {
      "epoch": 0.8832264110142003,
      "grad_norm": 0.13199752602727388,
      "learning_rate": 7.067801847596589e-06,
      "loss": 1.0855,
      "step": 12253
    },
    {
      "epoch": 0.8832984934765371,
      "grad_norm": 0.13046658661774957,
      "learning_rate": 7.059182967830458e-06,
      "loss": 0.9676,
      "step": 12254
    },
    {
      "epoch": 0.883370575938874,
      "grad_norm": 0.14585388342361236,
      "learning_rate": 7.050569154170317e-06,
      "loss": 1.0045,
      "step": 12255
    },
    {
      "epoch": 0.883442658401211,
      "grad_norm": 0.11495160700969831,
      "learning_rate": 7.041960407085724e-06,
      "loss": 1.0293,
      "step": 12256
    },
    {
      "epoch": 0.8835147408635479,
      "grad_norm": 0.10746788438796592,
      "learning_rate": 7.033356727045903e-06,
      "loss": 1.0601,
      "step": 12257
    },
    {
      "epoch": 0.8835868233258848,
      "grad_norm": 0.14022682358081362,
      "learning_rate": 7.024758114519847e-06,
      "loss": 1.0738,
      "step": 12258
    },
    {
      "epoch": 0.8836589057882217,
      "grad_norm": 0.1243025553322307,
      "learning_rate": 7.016164569976269e-06,
      "loss": 0.9996,
      "step": 12259
    },
    {
      "epoch": 0.8837309882505586,
      "grad_norm": 0.12051503554241444,
      "learning_rate": 7.00757609388355e-06,
      "loss": 1.0244,
      "step": 12260
    },
    {
      "epoch": 0.8838030707128955,
      "grad_norm": 0.13324978440576696,
      "learning_rate": 6.998992686709882e-06,
      "loss": 0.9971,
      "step": 12261
    },
    {
      "epoch": 0.8838751531752325,
      "grad_norm": 0.133999367001577,
      "learning_rate": 6.9904143489231244e-06,
      "loss": 0.9919,
      "step": 12262
    },
    {
      "epoch": 0.8839472356375694,
      "grad_norm": 0.3088496629707874,
      "learning_rate": 6.981841080990892e-06,
      "loss": 1.0283,
      "step": 12263
    },
    {
      "epoch": 0.8840193180999063,
      "grad_norm": 0.15289146414832558,
      "learning_rate": 6.973272883380444e-06,
      "loss": 0.979,
      "step": 12264
    },
    {
      "epoch": 0.8840914005622432,
      "grad_norm": 0.12387877462516896,
      "learning_rate": 6.964709756558885e-06,
      "loss": 0.9533,
      "step": 12265
    },
    {
      "epoch": 0.8841634830245801,
      "grad_norm": 0.12141214690978062,
      "learning_rate": 6.956151700992963e-06,
      "loss": 1.1193,
      "step": 12266
    },
    {
      "epoch": 0.8842355654869171,
      "grad_norm": 0.11617486922695296,
      "learning_rate": 6.947598717149184e-06,
      "loss": 0.9888,
      "step": 12267
    },
    {
      "epoch": 0.884307647949254,
      "grad_norm": 0.11414502825795284,
      "learning_rate": 6.93905080549373e-06,
      "loss": 1.0527,
      "step": 12268
    },
    {
      "epoch": 0.8843797304115909,
      "grad_norm": 0.13225483530372684,
      "learning_rate": 6.930507966492539e-06,
      "loss": 1.0405,
      "step": 12269
    },
    {
      "epoch": 0.8844518128739278,
      "grad_norm": 0.12027769117893301,
      "learning_rate": 6.921970200611317e-06,
      "loss": 1.0326,
      "step": 12270
    },
    {
      "epoch": 0.8845238953362646,
      "grad_norm": 0.1651672432396969,
      "learning_rate": 6.913437508315401e-06,
      "loss": 1.0086,
      "step": 12271
    },
    {
      "epoch": 0.8845959777986016,
      "grad_norm": 0.11770467598661637,
      "learning_rate": 6.9049098900699215e-06,
      "loss": 1.0804,
      "step": 12272
    },
    {
      "epoch": 0.8846680602609385,
      "grad_norm": 0.1003351027513488,
      "learning_rate": 6.896387346339683e-06,
      "loss": 1.0007,
      "step": 12273
    },
    {
      "epoch": 0.8847401427232754,
      "grad_norm": 0.14961377893044658,
      "learning_rate": 6.887869877589292e-06,
      "loss": 1.1713,
      "step": 12274
    },
    {
      "epoch": 0.8848122251856123,
      "grad_norm": 0.14146585471406392,
      "learning_rate": 6.879357484282989e-06,
      "loss": 1.0333,
      "step": 12275
    },
    {
      "epoch": 0.8848843076479492,
      "grad_norm": 0.11387344008739723,
      "learning_rate": 6.87085016688479e-06,
      "loss": 0.9823,
      "step": 12276
    },
    {
      "epoch": 0.8849563901102862,
      "grad_norm": 0.12279430397140269,
      "learning_rate": 6.862347925858392e-06,
      "loss": 1.0076,
      "step": 12277
    },
    {
      "epoch": 0.8850284725726231,
      "grad_norm": 0.13804632005605558,
      "learning_rate": 6.853850761667291e-06,
      "loss": 1.0879,
      "step": 12278
    },
    {
      "epoch": 0.88510055503496,
      "grad_norm": 0.12552213370341822,
      "learning_rate": 6.845358674774626e-06,
      "loss": 1.0751,
      "step": 12279
    },
    {
      "epoch": 0.8851726374972969,
      "grad_norm": 0.13216294632640274,
      "learning_rate": 6.836871665643297e-06,
      "loss": 1.0171,
      "step": 12280
    },
    {
      "epoch": 0.8852447199596338,
      "grad_norm": 0.11231906788464449,
      "learning_rate": 6.828389734735918e-06,
      "loss": 0.9696,
      "step": 12281
    },
    {
      "epoch": 0.8853168024219707,
      "grad_norm": 0.12880990427728345,
      "learning_rate": 6.819912882514834e-06,
      "loss": 1.0636,
      "step": 12282
    },
    {
      "epoch": 0.8853888848843077,
      "grad_norm": 0.12466760648540974,
      "learning_rate": 6.811441109442118e-06,
      "loss": 1.0334,
      "step": 12283
    },
    {
      "epoch": 0.8854609673466446,
      "grad_norm": 0.10693267930906648,
      "learning_rate": 6.802974415979546e-06,
      "loss": 0.9575,
      "step": 12284
    },
    {
      "epoch": 0.8855330498089815,
      "grad_norm": 0.11176571073553496,
      "learning_rate": 6.7945128025886375e-06,
      "loss": 1.0192,
      "step": 12285
    },
    {
      "epoch": 0.8856051322713184,
      "grad_norm": 0.11520301847350681,
      "learning_rate": 6.7860562697306114e-06,
      "loss": 1.0111,
      "step": 12286
    },
    {
      "epoch": 0.8856772147336553,
      "grad_norm": 0.12592255548013837,
      "learning_rate": 6.777604817866445e-06,
      "loss": 0.9414,
      "step": 12287
    },
    {
      "epoch": 0.8857492971959923,
      "grad_norm": 0.12881108070689928,
      "learning_rate": 6.769158447456802e-06,
      "loss": 1.0073,
      "step": 12288
    },
    {
      "epoch": 0.8858213796583291,
      "grad_norm": 0.17835371170229203,
      "learning_rate": 6.760717158962093e-06,
      "loss": 1.0301,
      "step": 12289
    },
    {
      "epoch": 0.885893462120666,
      "grad_norm": 0.136226263122762,
      "learning_rate": 6.752280952842438e-06,
      "loss": 0.9801,
      "step": 12290
    },
    {
      "epoch": 0.8859655445830029,
      "grad_norm": 0.16600585055936468,
      "learning_rate": 6.7438498295576915e-06,
      "loss": 1.1128,
      "step": 12291
    },
    {
      "epoch": 0.8860376270453398,
      "grad_norm": 0.9124817500732585,
      "learning_rate": 6.73542378956743e-06,
      "loss": 1.0601,
      "step": 12292
    },
    {
      "epoch": 0.8861097095076768,
      "grad_norm": 0.1222032386736554,
      "learning_rate": 6.727002833330931e-06,
      "loss": 1.0556,
      "step": 12293
    },
    {
      "epoch": 0.8861817919700137,
      "grad_norm": 0.1283210343825234,
      "learning_rate": 6.718586961307227e-06,
      "loss": 1.001,
      "step": 12294
    },
    {
      "epoch": 0.8862538744323506,
      "grad_norm": 0.1199062826111023,
      "learning_rate": 6.71017617395504e-06,
      "loss": 1.0666,
      "step": 12295
    },
    {
      "epoch": 0.8863259568946875,
      "grad_norm": 0.11610971416973928,
      "learning_rate": 6.70177047173286e-06,
      "loss": 1.0799,
      "step": 12296
    },
    {
      "epoch": 0.8863980393570244,
      "grad_norm": 0.28341941875188165,
      "learning_rate": 6.693369855098852e-06,
      "loss": 1.0731,
      "step": 12297
    },
    {
      "epoch": 0.8864701218193614,
      "grad_norm": 0.12236357644425451,
      "learning_rate": 6.6849743245109396e-06,
      "loss": 0.995,
      "step": 12298
    },
    {
      "epoch": 0.8865422042816983,
      "grad_norm": 0.11599527373015547,
      "learning_rate": 6.676583880426746e-06,
      "loss": 1.1004,
      "step": 12299
    },
    {
      "epoch": 0.8866142867440352,
      "grad_norm": 0.12279201637657415,
      "learning_rate": 6.6681985233036145e-06,
      "loss": 1.0651,
      "step": 12300
    },
    {
      "epoch": 0.8866863692063721,
      "grad_norm": 0.13169510946168742,
      "learning_rate": 6.659818253598648e-06,
      "loss": 1.0186,
      "step": 12301
    },
    {
      "epoch": 0.886758451668709,
      "grad_norm": 0.11258459776069162,
      "learning_rate": 6.651443071768626e-06,
      "loss": 1.0672,
      "step": 12302
    },
    {
      "epoch": 0.886830534131046,
      "grad_norm": 0.1252957533765872,
      "learning_rate": 6.6430729782700925e-06,
      "loss": 1.0262,
      "step": 12303
    },
    {
      "epoch": 0.8869026165933829,
      "grad_norm": 0.1418072085914191,
      "learning_rate": 6.63470797355924e-06,
      "loss": 1.0016,
      "step": 12304
    },
    {
      "epoch": 0.8869746990557198,
      "grad_norm": 0.14895699061742543,
      "learning_rate": 6.626348058092091e-06,
      "loss": 1.0603,
      "step": 12305
    },
    {
      "epoch": 0.8870467815180566,
      "grad_norm": 0.11403891707105687,
      "learning_rate": 6.617993232324315e-06,
      "loss": 1.0001,
      "step": 12306
    },
    {
      "epoch": 0.8871188639803935,
      "grad_norm": 0.12678856369101063,
      "learning_rate": 6.609643496711349e-06,
      "loss": 1.0686,
      "step": 12307
    },
    {
      "epoch": 0.8871909464427304,
      "grad_norm": 0.12021388685470197,
      "learning_rate": 6.601298851708271e-06,
      "loss": 1.028,
      "step": 12308
    },
    {
      "epoch": 0.8872630289050674,
      "grad_norm": 0.11438971946672348,
      "learning_rate": 6.5929592977699964e-06,
      "loss": 0.9935,
      "step": 12309
    },
    {
      "epoch": 0.8873351113674043,
      "grad_norm": 0.2076068133410004,
      "learning_rate": 6.584624835351072e-06,
      "loss": 1.0261,
      "step": 12310
    },
    {
      "epoch": 0.8874071938297412,
      "grad_norm": 0.1192371755519719,
      "learning_rate": 6.576295464905835e-06,
      "loss": 0.9998,
      "step": 12311
    },
    {
      "epoch": 0.8874792762920781,
      "grad_norm": 0.20908671202956383,
      "learning_rate": 6.5679711868882774e-06,
      "loss": 1.0098,
      "step": 12312
    },
    {
      "epoch": 0.887551358754415,
      "grad_norm": 0.122828125610288,
      "learning_rate": 6.559652001752137e-06,
      "loss": 1.1359,
      "step": 12313
    },
    {
      "epoch": 0.887623441216752,
      "grad_norm": 0.16148984444633213,
      "learning_rate": 6.551337909950916e-06,
      "loss": 1.0047,
      "step": 12314
    },
    {
      "epoch": 0.8876955236790889,
      "grad_norm": 0.10317126973262256,
      "learning_rate": 6.54302891193781e-06,
      "loss": 1.0622,
      "step": 12315
    },
    {
      "epoch": 0.8877676061414258,
      "grad_norm": 0.3221111476034956,
      "learning_rate": 6.534725008165699e-06,
      "loss": 1.0385,
      "step": 12316
    },
    {
      "epoch": 0.8878396886037627,
      "grad_norm": 0.11932678314136527,
      "learning_rate": 6.526426199087232e-06,
      "loss": 0.9799,
      "step": 12317
    },
    {
      "epoch": 0.8879117710660996,
      "grad_norm": 0.1263070231180344,
      "learning_rate": 6.518132485154793e-06,
      "loss": 1.0438,
      "step": 12318
    },
    {
      "epoch": 0.8879838535284366,
      "grad_norm": 0.17560928864039915,
      "learning_rate": 6.5098438668204535e-06,
      "loss": 1.0728,
      "step": 12319
    },
    {
      "epoch": 0.8880559359907735,
      "grad_norm": 0.13104684830130142,
      "learning_rate": 6.5015603445359955e-06,
      "loss": 1.0172,
      "step": 12320
    },
    {
      "epoch": 0.8881280184531104,
      "grad_norm": 0.1312311458434632,
      "learning_rate": 6.493281918752947e-06,
      "loss": 1.007,
      "step": 12321
    },
    {
      "epoch": 0.8882001009154473,
      "grad_norm": 0.11388547882375276,
      "learning_rate": 6.485008589922581e-06,
      "loss": 1.0136,
      "step": 12322
    },
    {
      "epoch": 0.8882721833777841,
      "grad_norm": 0.10912981311972063,
      "learning_rate": 6.4767403584958804e-06,
      "loss": 1.0078,
      "step": 12323
    },
    {
      "epoch": 0.888344265840121,
      "grad_norm": 0.15413389547537035,
      "learning_rate": 6.468477224923486e-06,
      "loss": 1.0153,
      "step": 12324
    },
    {
      "epoch": 0.888416348302458,
      "grad_norm": 0.1325518905312021,
      "learning_rate": 6.460219189655847e-06,
      "loss": 1.1086,
      "step": 12325
    },
    {
      "epoch": 0.8884884307647949,
      "grad_norm": 0.11870850616291559,
      "learning_rate": 6.451966253143082e-06,
      "loss": 1.0761,
      "step": 12326
    },
    {
      "epoch": 0.8885605132271318,
      "grad_norm": 0.12132137088703025,
      "learning_rate": 6.4437184158350765e-06,
      "loss": 1.0876,
      "step": 12327
    },
    {
      "epoch": 0.8886325956894687,
      "grad_norm": 0.15229864977035915,
      "learning_rate": 6.435475678181391e-06,
      "loss": 1.0048,
      "step": 12328
    },
    {
      "epoch": 0.8887046781518056,
      "grad_norm": 0.1371275883846029,
      "learning_rate": 6.4272380406313334e-06,
      "loss": 1.0,
      "step": 12329
    },
    {
      "epoch": 0.8887767606141426,
      "grad_norm": 0.11603281341616573,
      "learning_rate": 6.419005503633913e-06,
      "loss": 1.0129,
      "step": 12330
    },
    {
      "epoch": 0.8888488430764795,
      "grad_norm": 0.1487728842417015,
      "learning_rate": 6.410778067637924e-06,
      "loss": 1.0793,
      "step": 12331
    },
    {
      "epoch": 0.8889209255388164,
      "grad_norm": 0.11143906908136139,
      "learning_rate": 6.402555733091786e-06,
      "loss": 1.0623,
      "step": 12332
    },
    {
      "epoch": 0.8889930080011533,
      "grad_norm": 0.1322923135159633,
      "learning_rate": 6.394338500443697e-06,
      "loss": 1.0459,
      "step": 12333
    },
    {
      "epoch": 0.8890650904634902,
      "grad_norm": 0.12717396418411675,
      "learning_rate": 6.386126370141599e-06,
      "loss": 1.0251,
      "step": 12334
    },
    {
      "epoch": 0.8891371729258272,
      "grad_norm": 0.11324533980316322,
      "learning_rate": 6.377919342633098e-06,
      "loss": 1.0506,
      "step": 12335
    },
    {
      "epoch": 0.8892092553881641,
      "grad_norm": 0.16711776724928734,
      "learning_rate": 6.36971741836555e-06,
      "loss": 1.0502,
      "step": 12336
    },
    {
      "epoch": 0.889281337850501,
      "grad_norm": 0.1299215828762636,
      "learning_rate": 6.3615205977860526e-06,
      "loss": 1.0222,
      "step": 12337
    },
    {
      "epoch": 0.8893534203128379,
      "grad_norm": 0.21044338156849504,
      "learning_rate": 6.353328881341402e-06,
      "loss": 1.0803,
      "step": 12338
    },
    {
      "epoch": 0.8894255027751748,
      "grad_norm": 0.14710281993734786,
      "learning_rate": 6.345142269478111e-06,
      "loss": 1.0685,
      "step": 12339
    },
    {
      "epoch": 0.8894975852375118,
      "grad_norm": 0.11370327122667756,
      "learning_rate": 6.336960762642419e-06,
      "loss": 1.0139,
      "step": 12340
    },
    {
      "epoch": 0.8895696676998486,
      "grad_norm": 0.12337947371186765,
      "learning_rate": 6.328784361280304e-06,
      "loss": 0.9969,
      "step": 12341
    },
    {
      "epoch": 0.8896417501621855,
      "grad_norm": 0.13163889258997324,
      "learning_rate": 6.320613065837455e-06,
      "loss": 0.9908,
      "step": 12342
    },
    {
      "epoch": 0.8897138326245224,
      "grad_norm": 0.12350030406246988,
      "learning_rate": 6.312446876759259e-06,
      "loss": 1.0314,
      "step": 12343
    },
    {
      "epoch": 0.8897859150868593,
      "grad_norm": 0.10934942087406221,
      "learning_rate": 6.3042857944908715e-06,
      "loss": 1.0499,
      "step": 12344
    },
    {
      "epoch": 0.8898579975491963,
      "grad_norm": 0.13125761228754823,
      "learning_rate": 6.296129819477137e-06,
      "loss": 1.0717,
      "step": 12345
    },
    {
      "epoch": 0.8899300800115332,
      "grad_norm": 0.13497160986449702,
      "learning_rate": 6.287978952162621e-06,
      "loss": 1.039,
      "step": 12346
    },
    {
      "epoch": 0.8900021624738701,
      "grad_norm": 0.1647842973149842,
      "learning_rate": 6.2798331929916246e-06,
      "loss": 0.9039,
      "step": 12347
    },
    {
      "epoch": 0.890074244936207,
      "grad_norm": 0.12445042344940299,
      "learning_rate": 6.271692542408159e-06,
      "loss": 0.964,
      "step": 12348
    },
    {
      "epoch": 0.8901463273985439,
      "grad_norm": 0.23420589405854106,
      "learning_rate": 6.263557000855969e-06,
      "loss": 1.0283,
      "step": 12349
    },
    {
      "epoch": 0.8902184098608809,
      "grad_norm": 0.13233609010790776,
      "learning_rate": 6.255426568778522e-06,
      "loss": 0.985,
      "step": 12350
    },
    {
      "epoch": 0.8902904923232178,
      "grad_norm": 0.14079072439133258,
      "learning_rate": 6.247301246618975e-06,
      "loss": 1.0036,
      "step": 12351
    },
    {
      "epoch": 0.8903625747855547,
      "grad_norm": 0.12226564549038457,
      "learning_rate": 6.239181034820251e-06,
      "loss": 0.9849,
      "step": 12352
    },
    {
      "epoch": 0.8904346572478916,
      "grad_norm": 0.13833784539888208,
      "learning_rate": 6.231065933824976e-06,
      "loss": 1.0124,
      "step": 12353
    },
    {
      "epoch": 0.8905067397102285,
      "grad_norm": 0.11110100449428158,
      "learning_rate": 6.2229559440754815e-06,
      "loss": 1.0244,
      "step": 12354
    },
    {
      "epoch": 0.8905788221725655,
      "grad_norm": 0.10696590733893524,
      "learning_rate": 6.2148510660138496e-06,
      "loss": 1.0955,
      "step": 12355
    },
    {
      "epoch": 0.8906509046349024,
      "grad_norm": 0.12775469662782027,
      "learning_rate": 6.20675130008187e-06,
      "loss": 1.0329,
      "step": 12356
    },
    {
      "epoch": 0.8907229870972393,
      "grad_norm": 0.1144545202676461,
      "learning_rate": 6.198656646721012e-06,
      "loss": 1.0134,
      "step": 12357
    },
    {
      "epoch": 0.8907950695595761,
      "grad_norm": 0.11814667866168889,
      "learning_rate": 6.190567106372547e-06,
      "loss": 0.9893,
      "step": 12358
    },
    {
      "epoch": 0.890867152021913,
      "grad_norm": 0.1410453081265213,
      "learning_rate": 6.18248267947743e-06,
      "loss": 1.0177,
      "step": 12359
    },
    {
      "epoch": 0.8909392344842499,
      "grad_norm": 0.16217100518220226,
      "learning_rate": 6.1744033664763225e-06,
      "loss": 1.0574,
      "step": 12360
    },
    {
      "epoch": 0.8910113169465869,
      "grad_norm": 0.11079300468423307,
      "learning_rate": 6.166329167809604e-06,
      "loss": 1.015,
      "step": 12361
    },
    {
      "epoch": 0.8910833994089238,
      "grad_norm": 0.12165892767353047,
      "learning_rate": 6.158260083917411e-06,
      "loss": 0.9733,
      "step": 12362
    },
    {
      "epoch": 0.8911554818712607,
      "grad_norm": 0.1351647942668753,
      "learning_rate": 6.150196115239581e-06,
      "loss": 0.9725,
      "step": 12363
    },
    {
      "epoch": 0.8912275643335976,
      "grad_norm": 0.12696178598725041,
      "learning_rate": 6.142137262215686e-06,
      "loss": 0.9937,
      "step": 12364
    },
    {
      "epoch": 0.8912996467959345,
      "grad_norm": 0.11531609110445885,
      "learning_rate": 6.13408352528495e-06,
      "loss": 0.9789,
      "step": 12365
    },
    {
      "epoch": 0.8913717292582715,
      "grad_norm": 0.1298785315646552,
      "learning_rate": 6.126034904886424e-06,
      "loss": 1.0455,
      "step": 12366
    },
    {
      "epoch": 0.8914438117206084,
      "grad_norm": 0.1674609219799512,
      "learning_rate": 6.117991401458823e-06,
      "loss": 0.9956,
      "step": 12367
    },
    {
      "epoch": 0.8915158941829453,
      "grad_norm": 0.11455209861166431,
      "learning_rate": 6.109953015440595e-06,
      "loss": 1.0112,
      "step": 12368
    },
    {
      "epoch": 0.8915879766452822,
      "grad_norm": 0.12668819238309878,
      "learning_rate": 6.101919747269869e-06,
      "loss": 0.967,
      "step": 12369
    },
    {
      "epoch": 0.8916600591076191,
      "grad_norm": 0.10803241726502988,
      "learning_rate": 6.093891597384549e-06,
      "loss": 1.0342,
      "step": 12370
    },
    {
      "epoch": 0.8917321415699561,
      "grad_norm": 0.11750677695584143,
      "learning_rate": 6.085868566222252e-06,
      "loss": 1.0115,
      "step": 12371
    },
    {
      "epoch": 0.891804224032293,
      "grad_norm": 0.1361239417332395,
      "learning_rate": 6.077850654220318e-06,
      "loss": 1.0012,
      "step": 12372
    },
    {
      "epoch": 0.8918763064946299,
      "grad_norm": 0.13407075644536712,
      "learning_rate": 6.069837861815764e-06,
      "loss": 0.9704,
      "step": 12373
    },
    {
      "epoch": 0.8919483889569668,
      "grad_norm": 0.11263193572127254,
      "learning_rate": 6.061830189445339e-06,
      "loss": 0.9642,
      "step": 12374
    },
    {
      "epoch": 0.8920204714193036,
      "grad_norm": 0.12835385870426347,
      "learning_rate": 6.053827637545595e-06,
      "loss": 1.0507,
      "step": 12375
    },
    {
      "epoch": 0.8920925538816405,
      "grad_norm": 0.15768859678655076,
      "learning_rate": 6.045830206552716e-06,
      "loss": 1.0426,
      "step": 12376
    },
    {
      "epoch": 0.8921646363439775,
      "grad_norm": 0.10570885496054536,
      "learning_rate": 6.037837896902621e-06,
      "loss": 1.0134,
      "step": 12377
    },
    {
      "epoch": 0.8922367188063144,
      "grad_norm": 0.1711833432322193,
      "learning_rate": 6.029850709030971e-06,
      "loss": 1.103,
      "step": 12378
    },
    {
      "epoch": 0.8923088012686513,
      "grad_norm": 0.1201135816743099,
      "learning_rate": 6.021868643373119e-06,
      "loss": 0.9959,
      "step": 12379
    },
    {
      "epoch": 0.8923808837309882,
      "grad_norm": 0.11643380849117274,
      "learning_rate": 6.013891700364205e-06,
      "loss": 0.9946,
      "step": 12380
    },
    {
      "epoch": 0.8924529661933251,
      "grad_norm": 0.13713295812586584,
      "learning_rate": 6.005919880439004e-06,
      "loss": 1.0559,
      "step": 12381
    },
    {
      "epoch": 0.8925250486556621,
      "grad_norm": 0.4468568585215438,
      "learning_rate": 5.997953184032068e-06,
      "loss": 1.1151,
      "step": 12382
    },
    {
      "epoch": 0.892597131117999,
      "grad_norm": 0.11086858929088633,
      "learning_rate": 5.989991611577639e-06,
      "loss": 1.0296,
      "step": 12383
    },
    {
      "epoch": 0.8926692135803359,
      "grad_norm": 0.12108772981273293,
      "learning_rate": 5.982035163509725e-06,
      "loss": 1.02,
      "step": 12384
    },
    {
      "epoch": 0.8927412960426728,
      "grad_norm": 0.12919635734972776,
      "learning_rate": 5.974083840261991e-06,
      "loss": 1.0759,
      "step": 12385
    },
    {
      "epoch": 0.8928133785050097,
      "grad_norm": 0.12731809271192823,
      "learning_rate": 5.966137642267866e-06,
      "loss": 1.0208,
      "step": 12386
    },
    {
      "epoch": 0.8928854609673467,
      "grad_norm": 0.7731972439878627,
      "learning_rate": 5.958196569960483e-06,
      "loss": 1.0772,
      "step": 12387
    },
    {
      "epoch": 0.8929575434296836,
      "grad_norm": 0.13725371198150585,
      "learning_rate": 5.95026062377273e-06,
      "loss": 1.1033,
      "step": 12388
    },
    {
      "epoch": 0.8930296258920205,
      "grad_norm": 0.24673922852928543,
      "learning_rate": 5.942329804137159e-06,
      "loss": 1.1554,
      "step": 12389
    },
    {
      "epoch": 0.8931017083543574,
      "grad_norm": 0.11983450019098767,
      "learning_rate": 5.9344041114860805e-06,
      "loss": 1.0395,
      "step": 12390
    },
    {
      "epoch": 0.8931737908166943,
      "grad_norm": 0.11400576077539139,
      "learning_rate": 5.9264835462515046e-06,
      "loss": 1.0341,
      "step": 12391
    },
    {
      "epoch": 0.8932458732790313,
      "grad_norm": 0.11886772647427321,
      "learning_rate": 5.918568108865174e-06,
      "loss": 0.9941,
      "step": 12392
    },
    {
      "epoch": 0.8933179557413681,
      "grad_norm": 1.1002106435638324,
      "learning_rate": 5.910657799758568e-06,
      "loss": 0.9689,
      "step": 12393
    },
    {
      "epoch": 0.893390038203705,
      "grad_norm": 0.11605426330694256,
      "learning_rate": 5.902752619362861e-06,
      "loss": 1.009,
      "step": 12394
    },
    {
      "epoch": 0.8934621206660419,
      "grad_norm": 0.11734818890011285,
      "learning_rate": 5.894852568108955e-06,
      "loss": 1.0329,
      "step": 12395
    },
    {
      "epoch": 0.8935342031283788,
      "grad_norm": 0.12048010878670681,
      "learning_rate": 5.886957646427471e-06,
      "loss": 1.0292,
      "step": 12396
    },
    {
      "epoch": 0.8936062855907158,
      "grad_norm": 0.12894326887562252,
      "learning_rate": 5.879067854748743e-06,
      "loss": 0.9507,
      "step": 12397
    },
    {
      "epoch": 0.8936783680530527,
      "grad_norm": 0.11409206766070294,
      "learning_rate": 5.871183193502861e-06,
      "loss": 1.0155,
      "step": 12398
    },
    {
      "epoch": 0.8937504505153896,
      "grad_norm": 0.1394929973829705,
      "learning_rate": 5.8633036631195906e-06,
      "loss": 0.9849,
      "step": 12399
    },
    {
      "epoch": 0.8938225329777265,
      "grad_norm": 0.1234790101995403,
      "learning_rate": 5.855429264028444e-06,
      "loss": 0.9977,
      "step": 12400
    },
    {
      "epoch": 0.8938946154400634,
      "grad_norm": 0.18808393608417842,
      "learning_rate": 5.847559996658636e-06,
      "loss": 1.0275,
      "step": 12401
    },
    {
      "epoch": 0.8939666979024004,
      "grad_norm": 0.13339542441650878,
      "learning_rate": 5.839695861439132e-06,
      "loss": 1.0428,
      "step": 12402
    },
    {
      "epoch": 0.8940387803647373,
      "grad_norm": 0.12893364905044016,
      "learning_rate": 5.831836858798579e-06,
      "loss": 0.9945,
      "step": 12403
    },
    {
      "epoch": 0.8941108628270742,
      "grad_norm": 0.12136371658319212,
      "learning_rate": 5.823982989165367e-06,
      "loss": 1.0267,
      "step": 12404
    },
    {
      "epoch": 0.8941829452894111,
      "grad_norm": 0.13374828709362216,
      "learning_rate": 5.816134252967609e-06,
      "loss": 0.9711,
      "step": 12405
    },
    {
      "epoch": 0.894255027751748,
      "grad_norm": 0.11820602618454029,
      "learning_rate": 5.8082906506331306e-06,
      "loss": 1.0285,
      "step": 12406
    },
    {
      "epoch": 0.894327110214085,
      "grad_norm": 0.13476351330827202,
      "learning_rate": 5.8004521825894665e-06,
      "loss": 1.0487,
      "step": 12407
    },
    {
      "epoch": 0.8943991926764219,
      "grad_norm": 0.10927584223983328,
      "learning_rate": 5.792618849263898e-06,
      "loss": 1.0115,
      "step": 12408
    },
    {
      "epoch": 0.8944712751387588,
      "grad_norm": 0.1724847309079046,
      "learning_rate": 5.784790651083405e-06,
      "loss": 1.0512,
      "step": 12409
    },
    {
      "epoch": 0.8945433576010956,
      "grad_norm": 0.13137200462018617,
      "learning_rate": 5.776967588474691e-06,
      "loss": 1.0157,
      "step": 12410
    },
    {
      "epoch": 0.8946154400634325,
      "grad_norm": 0.13136784737335302,
      "learning_rate": 5.769149661864193e-06,
      "loss": 1.0468,
      "step": 12411
    },
    {
      "epoch": 0.8946875225257694,
      "grad_norm": 0.14025946038784964,
      "learning_rate": 5.761336871678047e-06,
      "loss": 1.0281,
      "step": 12412
    },
    {
      "epoch": 0.8947596049881064,
      "grad_norm": 0.14433207455089186,
      "learning_rate": 5.753529218342135e-06,
      "loss": 0.9615,
      "step": 12413
    },
    {
      "epoch": 0.8948316874504433,
      "grad_norm": 0.1552611190368088,
      "learning_rate": 5.745726702282006e-06,
      "loss": 0.9948,
      "step": 12414
    },
    {
      "epoch": 0.8949037699127802,
      "grad_norm": 0.14117710370232361,
      "learning_rate": 5.737929323923019e-06,
      "loss": 1.0359,
      "step": 12415
    },
    {
      "epoch": 0.8949758523751171,
      "grad_norm": 0.14447466524943298,
      "learning_rate": 5.730137083690168e-06,
      "loss": 1.0508,
      "step": 12416
    },
    {
      "epoch": 0.895047934837454,
      "grad_norm": 0.12005635582395234,
      "learning_rate": 5.7223499820082235e-06,
      "loss": 1.032,
      "step": 12417
    },
    {
      "epoch": 0.895120017299791,
      "grad_norm": 0.1409249724347752,
      "learning_rate": 5.714568019301603e-06,
      "loss": 1.0598,
      "step": 12418
    },
    {
      "epoch": 0.8951920997621279,
      "grad_norm": 0.13342628095404588,
      "learning_rate": 5.706791195994554e-06,
      "loss": 1.0446,
      "step": 12419
    },
    {
      "epoch": 0.8952641822244648,
      "grad_norm": 0.14417932274657536,
      "learning_rate": 5.69901951251095e-06,
      "loss": 1.0873,
      "step": 12420
    },
    {
      "epoch": 0.8953362646868017,
      "grad_norm": 0.11158751270545011,
      "learning_rate": 5.69125296927443e-06,
      "loss": 1.0256,
      "step": 12421
    },
    {
      "epoch": 0.8954083471491386,
      "grad_norm": 0.14561938268610652,
      "learning_rate": 5.683491566708332e-06,
      "loss": 1.0451,
      "step": 12422
    },
    {
      "epoch": 0.8954804296114756,
      "grad_norm": 0.14593715686045064,
      "learning_rate": 5.6757353052356964e-06,
      "loss": 1.0954,
      "step": 12423
    },
    {
      "epoch": 0.8955525120738125,
      "grad_norm": 0.12536207873817606,
      "learning_rate": 5.667984185279374e-06,
      "loss": 0.9996,
      "step": 12424
    },
    {
      "epoch": 0.8956245945361494,
      "grad_norm": 0.12214340602864801,
      "learning_rate": 5.660238207261814e-06,
      "loss": 0.9899,
      "step": 12425
    },
    {
      "epoch": 0.8956966769984863,
      "grad_norm": 0.14726855100328504,
      "learning_rate": 5.652497371605259e-06,
      "loss": 1.0095,
      "step": 12426
    },
    {
      "epoch": 0.8957687594608231,
      "grad_norm": 0.12632447690057377,
      "learning_rate": 5.644761678731647e-06,
      "loss": 1.0185,
      "step": 12427
    },
    {
      "epoch": 0.89584084192316,
      "grad_norm": 0.27686545966717424,
      "learning_rate": 5.6370311290626645e-06,
      "loss": 1.1118,
      "step": 12428
    },
    {
      "epoch": 0.895912924385497,
      "grad_norm": 0.1552768218750538,
      "learning_rate": 5.629305723019673e-06,
      "loss": 1.1465,
      "step": 12429
    },
    {
      "epoch": 0.8959850068478339,
      "grad_norm": 0.12933175979433037,
      "learning_rate": 5.6215854610237925e-06,
      "loss": 1.0328,
      "step": 12430
    },
    {
      "epoch": 0.8960570893101708,
      "grad_norm": 0.11569825565885707,
      "learning_rate": 5.6138703434958084e-06,
      "loss": 1.0143,
      "step": 12431
    },
    {
      "epoch": 0.8961291717725077,
      "grad_norm": 0.11790477798799606,
      "learning_rate": 5.606160370856328e-06,
      "loss": 1.0796,
      "step": 12432
    },
    {
      "epoch": 0.8962012542348446,
      "grad_norm": 0.11399301371052607,
      "learning_rate": 5.598455543525571e-06,
      "loss": 0.9449,
      "step": 12433
    },
    {
      "epoch": 0.8962733366971816,
      "grad_norm": 0.18075848865876784,
      "learning_rate": 5.590755861923513e-06,
      "loss": 1.0442,
      "step": 12434
    },
    {
      "epoch": 0.8963454191595185,
      "grad_norm": 0.1683808594549056,
      "learning_rate": 5.583061326469885e-06,
      "loss": 1.0194,
      "step": 12435
    },
    {
      "epoch": 0.8964175016218554,
      "grad_norm": 0.12085599901662737,
      "learning_rate": 5.5753719375840845e-06,
      "loss": 1.0888,
      "step": 12436
    },
    {
      "epoch": 0.8964895840841923,
      "grad_norm": 0.1227160811936418,
      "learning_rate": 5.567687695685264e-06,
      "loss": 1.0757,
      "step": 12437
    },
    {
      "epoch": 0.8965616665465292,
      "grad_norm": 0.12127614757173333,
      "learning_rate": 5.560008601192279e-06,
      "loss": 1.0168,
      "step": 12438
    },
    {
      "epoch": 0.8966337490088662,
      "grad_norm": 0.17769878837618608,
      "learning_rate": 5.552334654523705e-06,
      "loss": 1.0184,
      "step": 12439
    },
    {
      "epoch": 0.8967058314712031,
      "grad_norm": 0.12586968626617823,
      "learning_rate": 5.544665856097842e-06,
      "loss": 1.0655,
      "step": 12440
    },
    {
      "epoch": 0.89677791393354,
      "grad_norm": 0.11274567258091042,
      "learning_rate": 5.53700220633272e-06,
      "loss": 1.0521,
      "step": 12441
    },
    {
      "epoch": 0.8968499963958769,
      "grad_norm": 0.12395433038921838,
      "learning_rate": 5.529343705646073e-06,
      "loss": 1.0505,
      "step": 12442
    },
    {
      "epoch": 0.8969220788582138,
      "grad_norm": 0.12415151525854642,
      "learning_rate": 5.521690354455344e-06,
      "loss": 0.9777,
      "step": 12443
    },
    {
      "epoch": 0.8969941613205508,
      "grad_norm": 0.13597130856610246,
      "learning_rate": 5.514042153177712e-06,
      "loss": 1.0559,
      "step": 12444
    },
    {
      "epoch": 0.8970662437828876,
      "grad_norm": 0.1191006838766738,
      "learning_rate": 5.506399102230086e-06,
      "loss": 1.0669,
      "step": 12445
    },
    {
      "epoch": 0.8971383262452245,
      "grad_norm": 0.27303500523626356,
      "learning_rate": 5.498761202029079e-06,
      "loss": 0.9867,
      "step": 12446
    },
    {
      "epoch": 0.8972104087075614,
      "grad_norm": 0.10906815406440466,
      "learning_rate": 5.4911284529910126e-06,
      "loss": 1.0568,
      "step": 12447
    },
    {
      "epoch": 0.8972824911698983,
      "grad_norm": 0.12912303362316183,
      "learning_rate": 5.483500855531942e-06,
      "loss": 1.0236,
      "step": 12448
    },
    {
      "epoch": 0.8973545736322353,
      "grad_norm": 0.11290363787408597,
      "learning_rate": 5.475878410067647e-06,
      "loss": 1.1044,
      "step": 12449
    },
    {
      "epoch": 0.8974266560945722,
      "grad_norm": 0.11872218887393461,
      "learning_rate": 5.468261117013618e-06,
      "loss": 1.0015,
      "step": 12450
    },
    {
      "epoch": 0.8974987385569091,
      "grad_norm": 0.15441305742467548,
      "learning_rate": 5.460648976785054e-06,
      "loss": 1.0019,
      "step": 12451
    },
    {
      "epoch": 0.897570821019246,
      "grad_norm": 0.14099238953709567,
      "learning_rate": 5.453041989796903e-06,
      "loss": 1.0244,
      "step": 12452
    },
    {
      "epoch": 0.8976429034815829,
      "grad_norm": 0.11996064850098087,
      "learning_rate": 5.445440156463799e-06,
      "loss": 1.0185,
      "step": 12453
    },
    {
      "epoch": 0.8977149859439199,
      "grad_norm": 0.12242866645303875,
      "learning_rate": 5.437843477200122e-06,
      "loss": 1.0825,
      "step": 12454
    },
    {
      "epoch": 0.8977870684062568,
      "grad_norm": 0.15622767754723063,
      "learning_rate": 5.430251952419951e-06,
      "loss": 1.0915,
      "step": 12455
    },
    {
      "epoch": 0.8978591508685937,
      "grad_norm": 0.11286878352690073,
      "learning_rate": 5.42266558253709e-06,
      "loss": 0.9673,
      "step": 12456
    },
    {
      "epoch": 0.8979312333309306,
      "grad_norm": 0.10952194527516622,
      "learning_rate": 5.4150843679650845e-06,
      "loss": 1.0289,
      "step": 12457
    },
    {
      "epoch": 0.8980033157932675,
      "grad_norm": 0.37514161492221737,
      "learning_rate": 5.407508309117138e-06,
      "loss": 1.0931,
      "step": 12458
    },
    {
      "epoch": 0.8980753982556045,
      "grad_norm": 0.13201668151520196,
      "learning_rate": 5.399937406406241e-06,
      "loss": 1.0871,
      "step": 12459
    },
    {
      "epoch": 0.8981474807179414,
      "grad_norm": 0.11381545809104407,
      "learning_rate": 5.392371660245077e-06,
      "loss": 1.0433,
      "step": 12460
    },
    {
      "epoch": 0.8982195631802783,
      "grad_norm": 0.11765374376130874,
      "learning_rate": 5.384811071046059e-06,
      "loss": 1.0293,
      "step": 12461
    },
    {
      "epoch": 0.8982916456426151,
      "grad_norm": 0.12401985972758175,
      "learning_rate": 5.377255639221257e-06,
      "loss": 0.9497,
      "step": 12462
    },
    {
      "epoch": 0.898363728104952,
      "grad_norm": 0.11981697443090679,
      "learning_rate": 5.369705365182554e-06,
      "loss": 1.0541,
      "step": 12463
    },
    {
      "epoch": 0.8984358105672889,
      "grad_norm": 0.14901461321119222,
      "learning_rate": 5.362160249341497e-06,
      "loss": 1.05,
      "step": 12464
    },
    {
      "epoch": 0.8985078930296259,
      "grad_norm": 0.15695381242340847,
      "learning_rate": 5.354620292109381e-06,
      "loss": 1.0579,
      "step": 12465
    },
    {
      "epoch": 0.8985799754919628,
      "grad_norm": 0.12368714180433936,
      "learning_rate": 5.3470854938971635e-06,
      "loss": 1.0233,
      "step": 12466
    },
    {
      "epoch": 0.8986520579542997,
      "grad_norm": 0.12636261983500127,
      "learning_rate": 5.339555855115563e-06,
      "loss": 1.0072,
      "step": 12467
    },
    {
      "epoch": 0.8987241404166366,
      "grad_norm": 0.14113095942142886,
      "learning_rate": 5.332031376175029e-06,
      "loss": 1.0851,
      "step": 12468
    },
    {
      "epoch": 0.8987962228789735,
      "grad_norm": 0.13433279848328245,
      "learning_rate": 5.324512057485731e-06,
      "loss": 1.0519,
      "step": 12469
    },
    {
      "epoch": 0.8988683053413105,
      "grad_norm": 0.11477120449130807,
      "learning_rate": 5.31699789945751e-06,
      "loss": 0.9723,
      "step": 12470
    },
    {
      "epoch": 0.8989403878036474,
      "grad_norm": 0.12048237756989359,
      "learning_rate": 5.309488902499937e-06,
      "loss": 1.0213,
      "step": 12471
    },
    {
      "epoch": 0.8990124702659843,
      "grad_norm": 0.20230565315880122,
      "learning_rate": 5.301985067022364e-06,
      "loss": 0.9697,
      "step": 12472
    },
    {
      "epoch": 0.8990845527283212,
      "grad_norm": 0.1715556853344538,
      "learning_rate": 5.294486393433806e-06,
      "loss": 0.9452,
      "step": 12473
    },
    {
      "epoch": 0.8991566351906581,
      "grad_norm": 0.20696153785829846,
      "learning_rate": 5.286992882142982e-06,
      "loss": 1.0111,
      "step": 12474
    },
    {
      "epoch": 0.8992287176529951,
      "grad_norm": 0.11773869760193674,
      "learning_rate": 5.279504533558388e-06,
      "loss": 0.9411,
      "step": 12475
    },
    {
      "epoch": 0.899300800115332,
      "grad_norm": 0.21626328900646108,
      "learning_rate": 5.272021348088163e-06,
      "loss": 1.0351,
      "step": 12476
    },
    {
      "epoch": 0.8993728825776689,
      "grad_norm": 0.6046576735042192,
      "learning_rate": 5.2645433261402695e-06,
      "loss": 1.0163,
      "step": 12477
    },
    {
      "epoch": 0.8994449650400058,
      "grad_norm": 0.13626587316111927,
      "learning_rate": 5.257070468122272e-06,
      "loss": 1.1221,
      "step": 12478
    },
    {
      "epoch": 0.8995170475023426,
      "grad_norm": 0.12712163104271995,
      "learning_rate": 5.249602774441542e-06,
      "loss": 1.0482,
      "step": 12479
    },
    {
      "epoch": 0.8995891299646795,
      "grad_norm": 0.12707937912329678,
      "learning_rate": 5.2421402455050895e-06,
      "loss": 1.0392,
      "step": 12480
    },
    {
      "epoch": 0.8996612124270165,
      "grad_norm": 0.13095301795504496,
      "learning_rate": 5.2346828817197655e-06,
      "loss": 0.9807,
      "step": 12481
    },
    {
      "epoch": 0.8997332948893534,
      "grad_norm": 0.13945028824059022,
      "learning_rate": 5.227230683492002e-06,
      "loss": 1.0624,
      "step": 12482
    },
    {
      "epoch": 0.8998053773516903,
      "grad_norm": 0.11083871855244752,
      "learning_rate": 5.219783651228017e-06,
      "loss": 0.9495,
      "step": 12483
    },
    {
      "epoch": 0.8998774598140272,
      "grad_norm": 0.11284347850653895,
      "learning_rate": 5.212341785333752e-06,
      "loss": 0.9484,
      "step": 12484
    },
    {
      "epoch": 0.8999495422763641,
      "grad_norm": 0.21277317707399948,
      "learning_rate": 5.204905086214873e-06,
      "loss": 1.0453,
      "step": 12485
    },
    {
      "epoch": 0.9000216247387011,
      "grad_norm": 0.18493297706757292,
      "learning_rate": 5.197473554276711e-06,
      "loss": 1.0079,
      "step": 12486
    },
    {
      "epoch": 0.900093707201038,
      "grad_norm": 0.16519681263341637,
      "learning_rate": 5.190047189924374e-06,
      "loss": 0.9796,
      "step": 12487
    },
    {
      "epoch": 0.9001657896633749,
      "grad_norm": 0.1281730848775244,
      "learning_rate": 5.182625993562662e-06,
      "loss": 0.9306,
      "step": 12488
    },
    {
      "epoch": 0.9002378721257118,
      "grad_norm": 0.1278055543036163,
      "learning_rate": 5.175209965596084e-06,
      "loss": 1.006,
      "step": 12489
    },
    {
      "epoch": 0.9003099545880487,
      "grad_norm": 0.29388841734568105,
      "learning_rate": 5.167799106428894e-06,
      "loss": 1.0602,
      "step": 12490
    },
    {
      "epoch": 0.9003820370503857,
      "grad_norm": 0.14920816070279644,
      "learning_rate": 5.160393416465048e-06,
      "loss": 1.0257,
      "step": 12491
    },
    {
      "epoch": 0.9004541195127226,
      "grad_norm": 0.13328440847826997,
      "learning_rate": 5.152992896108222e-06,
      "loss": 1.0411,
      "step": 12492
    },
    {
      "epoch": 0.9005262019750595,
      "grad_norm": 0.12558956697878393,
      "learning_rate": 5.145597545761805e-06,
      "loss": 1.005,
      "step": 12493
    },
    {
      "epoch": 0.9005982844373964,
      "grad_norm": 0.11453709400734226,
      "learning_rate": 5.138207365828918e-06,
      "loss": 1.0186,
      "step": 12494
    },
    {
      "epoch": 0.9006703668997333,
      "grad_norm": 0.11979123998750535,
      "learning_rate": 5.130822356712384e-06,
      "loss": 1.0842,
      "step": 12495
    },
    {
      "epoch": 0.9007424493620703,
      "grad_norm": 0.1384752127537822,
      "learning_rate": 5.123442518814759e-06,
      "loss": 1.0588,
      "step": 12496
    },
    {
      "epoch": 0.9008145318244071,
      "grad_norm": 0.13164353341680232,
      "learning_rate": 5.116067852538298e-06,
      "loss": 0.9936,
      "step": 12497
    },
    {
      "epoch": 0.900886614286744,
      "grad_norm": 0.12359876143666278,
      "learning_rate": 5.108698358285002e-06,
      "loss": 1.0274,
      "step": 12498
    },
    {
      "epoch": 0.9009586967490809,
      "grad_norm": 0.15919724377156863,
      "learning_rate": 5.101334036456562e-06,
      "loss": 1.0008,
      "step": 12499
    },
    {
      "epoch": 0.9010307792114178,
      "grad_norm": 0.1269016327290945,
      "learning_rate": 5.093974887454411e-06,
      "loss": 1.0767,
      "step": 12500
    },
    {
      "epoch": 0.9011028616737548,
      "grad_norm": 0.24627369277816258,
      "learning_rate": 5.0866209116796845e-06,
      "loss": 1.035,
      "step": 12501
    },
    {
      "epoch": 0.9011749441360917,
      "grad_norm": 0.1348396417724481,
      "learning_rate": 5.079272109533229e-06,
      "loss": 1.1107,
      "step": 12502
    },
    {
      "epoch": 0.9012470265984286,
      "grad_norm": 0.16824740277192912,
      "learning_rate": 5.071928481415633e-06,
      "loss": 1.0295,
      "step": 12503
    },
    {
      "epoch": 0.9013191090607655,
      "grad_norm": 0.12375604380351847,
      "learning_rate": 5.06459002772719e-06,
      "loss": 1.0963,
      "step": 12504
    },
    {
      "epoch": 0.9013911915231024,
      "grad_norm": 0.1229505447842981,
      "learning_rate": 5.0572567488679005e-06,
      "loss": 1.0688,
      "step": 12505
    },
    {
      "epoch": 0.9014632739854394,
      "grad_norm": 0.12085401087088296,
      "learning_rate": 5.049928645237501e-06,
      "loss": 1.0164,
      "step": 12506
    },
    {
      "epoch": 0.9015353564477763,
      "grad_norm": 0.12707912464467827,
      "learning_rate": 5.042605717235438e-06,
      "loss": 0.9737,
      "step": 12507
    },
    {
      "epoch": 0.9016074389101132,
      "grad_norm": 0.12498976511411161,
      "learning_rate": 5.035287965260871e-06,
      "loss": 1.0508,
      "step": 12508
    },
    {
      "epoch": 0.9016795213724501,
      "grad_norm": 0.1280783639857661,
      "learning_rate": 5.027975389712702e-06,
      "loss": 1.1069,
      "step": 12509
    },
    {
      "epoch": 0.901751603834787,
      "grad_norm": 0.10951871491546074,
      "learning_rate": 5.020667990989525e-06,
      "loss": 0.9899,
      "step": 12510
    },
    {
      "epoch": 0.901823686297124,
      "grad_norm": 0.16167132748552165,
      "learning_rate": 5.01336576948962e-06,
      "loss": 1.0468,
      "step": 12511
    },
    {
      "epoch": 0.9018957687594609,
      "grad_norm": 0.11615968355418231,
      "learning_rate": 5.006068725611079e-06,
      "loss": 1.0988,
      "step": 12512
    },
    {
      "epoch": 0.9019678512217978,
      "grad_norm": 0.14930640565470368,
      "learning_rate": 4.998776859751619e-06,
      "loss": 0.9539,
      "step": 12513
    },
    {
      "epoch": 0.9020399336841346,
      "grad_norm": 0.10850689435857726,
      "learning_rate": 4.9914901723087544e-06,
      "loss": 0.9847,
      "step": 12514
    },
    {
      "epoch": 0.9021120161464715,
      "grad_norm": 0.12785836061365022,
      "learning_rate": 4.984208663679602e-06,
      "loss": 1.0149,
      "step": 12515
    },
    {
      "epoch": 0.9021840986088084,
      "grad_norm": 0.12061014076636185,
      "learning_rate": 4.9769323342611414e-06,
      "loss": 1.0762,
      "step": 12516
    },
    {
      "epoch": 0.9022561810711454,
      "grad_norm": 0.24968809142448659,
      "learning_rate": 4.969661184449958e-06,
      "loss": 1.0244,
      "step": 12517
    },
    {
      "epoch": 0.9023282635334823,
      "grad_norm": 0.1331678817198472,
      "learning_rate": 4.962395214642424e-06,
      "loss": 1.0827,
      "step": 12518
    },
    {
      "epoch": 0.9024003459958192,
      "grad_norm": 0.12238925177325902,
      "learning_rate": 4.955134425234576e-06,
      "loss": 1.0696,
      "step": 12519
    },
    {
      "epoch": 0.9024724284581561,
      "grad_norm": 0.1255976105442831,
      "learning_rate": 4.947878816622175e-06,
      "loss": 0.9749,
      "step": 12520
    },
    {
      "epoch": 0.902544510920493,
      "grad_norm": 0.14589521277717257,
      "learning_rate": 4.940628389200752e-06,
      "loss": 1.0478,
      "step": 12521
    },
    {
      "epoch": 0.90261659338283,
      "grad_norm": 0.11681667436746743,
      "learning_rate": 4.933383143365522e-06,
      "loss": 1.0697,
      "step": 12522
    },
    {
      "epoch": 0.9026886758451669,
      "grad_norm": 0.13596427536962685,
      "learning_rate": 4.926143079511391e-06,
      "loss": 1.045,
      "step": 12523
    },
    {
      "epoch": 0.9027607583075038,
      "grad_norm": 0.12862687665774694,
      "learning_rate": 4.918908198033001e-06,
      "loss": 1.1197,
      "step": 12524
    },
    {
      "epoch": 0.9028328407698407,
      "grad_norm": 0.2745425508868042,
      "learning_rate": 4.911678499324746e-06,
      "loss": 1.0416,
      "step": 12525
    },
    {
      "epoch": 0.9029049232321776,
      "grad_norm": 0.1327304021523939,
      "learning_rate": 4.90445398378071e-06,
      "loss": 1.1255,
      "step": 12526
    },
    {
      "epoch": 0.9029770056945146,
      "grad_norm": 0.11893545504544861,
      "learning_rate": 4.8972346517946684e-06,
      "loss": 1.0323,
      "step": 12527
    },
    {
      "epoch": 0.9030490881568515,
      "grad_norm": 0.11606480559651686,
      "learning_rate": 4.89002050376014e-06,
      "loss": 0.9516,
      "step": 12528
    },
    {
      "epoch": 0.9031211706191884,
      "grad_norm": 0.12695101296707437,
      "learning_rate": 4.882811540070387e-06,
      "loss": 1.1051,
      "step": 12529
    },
    {
      "epoch": 0.9031932530815253,
      "grad_norm": 0.11090319223168096,
      "learning_rate": 4.875607761118362e-06,
      "loss": 0.9828,
      "step": 12530
    },
    {
      "epoch": 0.9032653355438621,
      "grad_norm": 0.19551929282376157,
      "learning_rate": 4.868409167296706e-06,
      "loss": 1.0709,
      "step": 12531
    },
    {
      "epoch": 0.903337418006199,
      "grad_norm": 0.10851177507204643,
      "learning_rate": 4.861215758997817e-06,
      "loss": 1.0784,
      "step": 12532
    },
    {
      "epoch": 0.903409500468536,
      "grad_norm": 0.1392636699551725,
      "learning_rate": 4.854027536613792e-06,
      "loss": 1.0358,
      "step": 12533
    },
    {
      "epoch": 0.9034815829308729,
      "grad_norm": 0.13032882318383368,
      "learning_rate": 4.846844500536496e-06,
      "loss": 1.0879,
      "step": 12534
    },
    {
      "epoch": 0.9035536653932098,
      "grad_norm": 0.15080474115107834,
      "learning_rate": 4.839666651157426e-06,
      "loss": 1.077,
      "step": 12535
    },
    {
      "epoch": 0.9036257478555467,
      "grad_norm": 0.12143015925769154,
      "learning_rate": 4.832493988867859e-06,
      "loss": 1.0349,
      "step": 12536
    },
    {
      "epoch": 0.9036978303178836,
      "grad_norm": 0.13414582760650376,
      "learning_rate": 4.825326514058737e-06,
      "loss": 1.1394,
      "step": 12537
    },
    {
      "epoch": 0.9037699127802206,
      "grad_norm": 0.13826141367362288,
      "learning_rate": 4.818164227120814e-06,
      "loss": 0.9457,
      "step": 12538
    },
    {
      "epoch": 0.9038419952425575,
      "grad_norm": 0.11094259908078938,
      "learning_rate": 4.811007128444445e-06,
      "loss": 1.0041,
      "step": 12539
    },
    {
      "epoch": 0.9039140777048944,
      "grad_norm": 0.11920765417077457,
      "learning_rate": 4.8038552184197724e-06,
      "loss": 1.0847,
      "step": 12540
    },
    {
      "epoch": 0.9039861601672313,
      "grad_norm": 0.11132371479663011,
      "learning_rate": 4.796708497436641e-06,
      "loss": 1.0121,
      "step": 12541
    },
    {
      "epoch": 0.9040582426295682,
      "grad_norm": 0.23346781833002142,
      "learning_rate": 4.789566965884607e-06,
      "loss": 1.0472,
      "step": 12542
    },
    {
      "epoch": 0.9041303250919052,
      "grad_norm": 0.11171680954565512,
      "learning_rate": 4.782430624152945e-06,
      "loss": 1.0399,
      "step": 12543
    },
    {
      "epoch": 0.9042024075542421,
      "grad_norm": 0.11637671116139932,
      "learning_rate": 4.775299472630656e-06,
      "loss": 0.9771,
      "step": 12544
    },
    {
      "epoch": 0.904274490016579,
      "grad_norm": 0.1206592873050375,
      "learning_rate": 4.768173511706464e-06,
      "loss": 1.0332,
      "step": 12545
    },
    {
      "epoch": 0.9043465724789159,
      "grad_norm": 0.12354848331712753,
      "learning_rate": 4.7610527417687675e-06,
      "loss": 1.0442,
      "step": 12546
    },
    {
      "epoch": 0.9044186549412528,
      "grad_norm": 0.11376945772836103,
      "learning_rate": 4.753937163205735e-06,
      "loss": 1.0258,
      "step": 12547
    },
    {
      "epoch": 0.9044907374035898,
      "grad_norm": 0.11060696732610253,
      "learning_rate": 4.746826776405222e-06,
      "loss": 1.0664,
      "step": 12548
    },
    {
      "epoch": 0.9045628198659266,
      "grad_norm": 0.12128749843947931,
      "learning_rate": 4.739721581754808e-06,
      "loss": 1.0165,
      "step": 12549
    },
    {
      "epoch": 0.9046349023282635,
      "grad_norm": 0.16439491938395936,
      "learning_rate": 4.732621579641794e-06,
      "loss": 1.0042,
      "step": 12550
    },
    {
      "epoch": 0.9047069847906004,
      "grad_norm": 0.1252051786809114,
      "learning_rate": 4.7255267704531925e-06,
      "loss": 1.0427,
      "step": 12551
    },
    {
      "epoch": 0.9047790672529373,
      "grad_norm": 0.14770293216461863,
      "learning_rate": 4.718437154575728e-06,
      "loss": 1.0317,
      "step": 12552
    },
    {
      "epoch": 0.9048511497152742,
      "grad_norm": 0.13157088587830276,
      "learning_rate": 4.711352732395846e-06,
      "loss": 1.0005,
      "step": 12553
    },
    {
      "epoch": 0.9049232321776112,
      "grad_norm": 0.11671990084225577,
      "learning_rate": 4.704273504299727e-06,
      "loss": 1.0906,
      "step": 12554
    },
    {
      "epoch": 0.9049953146399481,
      "grad_norm": 0.11227910108131818,
      "learning_rate": 4.697199470673242e-06,
      "loss": 1.0569,
      "step": 12555
    },
    {
      "epoch": 0.905067397102285,
      "grad_norm": 0.1402405181727123,
      "learning_rate": 4.690130631901979e-06,
      "loss": 1.0493,
      "step": 12556
    },
    {
      "epoch": 0.9051394795646219,
      "grad_norm": 0.18347061622047844,
      "learning_rate": 4.683066988371265e-06,
      "loss": 1.0905,
      "step": 12557
    },
    {
      "epoch": 0.9052115620269588,
      "grad_norm": 0.12186076937452175,
      "learning_rate": 4.6760085404661365e-06,
      "loss": 0.9171,
      "step": 12558
    },
    {
      "epoch": 0.9052836444892958,
      "grad_norm": 0.12256797614708509,
      "learning_rate": 4.668955288571331e-06,
      "loss": 0.9883,
      "step": 12559
    },
    {
      "epoch": 0.9053557269516327,
      "grad_norm": 0.11104833788578582,
      "learning_rate": 4.6619072330713164e-06,
      "loss": 1.0355,
      "step": 12560
    },
    {
      "epoch": 0.9054278094139696,
      "grad_norm": 0.13693370285481446,
      "learning_rate": 4.654864374350287e-06,
      "loss": 1.0484,
      "step": 12561
    },
    {
      "epoch": 0.9054998918763065,
      "grad_norm": 0.12231731457537744,
      "learning_rate": 4.647826712792125e-06,
      "loss": 1.1549,
      "step": 12562
    },
    {
      "epoch": 0.9055719743386434,
      "grad_norm": 0.12441634663176417,
      "learning_rate": 4.640794248780455e-06,
      "loss": 1.0703,
      "step": 12563
    },
    {
      "epoch": 0.9056440568009804,
      "grad_norm": 0.10849668695115475,
      "learning_rate": 4.633766982698595e-06,
      "loss": 0.9871,
      "step": 12564
    },
    {
      "epoch": 0.9057161392633173,
      "grad_norm": 0.1332998449849562,
      "learning_rate": 4.626744914929615e-06,
      "loss": 1.0297,
      "step": 12565
    },
    {
      "epoch": 0.9057882217256541,
      "grad_norm": 0.15722894205646284,
      "learning_rate": 4.619728045856264e-06,
      "loss": 1.0679,
      "step": 12566
    },
    {
      "epoch": 0.905860304187991,
      "grad_norm": 0.12325808673113811,
      "learning_rate": 4.6127163758610485e-06,
      "loss": 1.1258,
      "step": 12567
    },
    {
      "epoch": 0.9059323866503279,
      "grad_norm": 0.1895827520590206,
      "learning_rate": 4.605709905326117e-06,
      "loss": 0.9767,
      "step": 12568
    },
    {
      "epoch": 0.9060044691126649,
      "grad_norm": 0.12584936563382784,
      "learning_rate": 4.598708634633442e-06,
      "loss": 1.0446,
      "step": 12569
    },
    {
      "epoch": 0.9060765515750018,
      "grad_norm": 0.140750132758335,
      "learning_rate": 4.591712564164619e-06,
      "loss": 0.9764,
      "step": 12570
    },
    {
      "epoch": 0.9061486340373387,
      "grad_norm": 0.13487245985185373,
      "learning_rate": 4.584721694301031e-06,
      "loss": 1.0622,
      "step": 12571
    },
    {
      "epoch": 0.9062207164996756,
      "grad_norm": 0.16107719737550033,
      "learning_rate": 4.577736025423684e-06,
      "loss": 1.0733,
      "step": 12572
    },
    {
      "epoch": 0.9062927989620125,
      "grad_norm": 0.11909861393601494,
      "learning_rate": 4.57075555791342e-06,
      "loss": 1.0006,
      "step": 12573
    },
    {
      "epoch": 0.9063648814243495,
      "grad_norm": 0.13230818324379587,
      "learning_rate": 4.563780292150721e-06,
      "loss": 1.1371,
      "step": 12574
    },
    {
      "epoch": 0.9064369638866864,
      "grad_norm": 0.12639943628529735,
      "learning_rate": 4.556810228515773e-06,
      "loss": 1.0392,
      "step": 12575
    },
    {
      "epoch": 0.9065090463490233,
      "grad_norm": 0.12876181526878902,
      "learning_rate": 4.549845367388539e-06,
      "loss": 1.0242,
      "step": 12576
    },
    {
      "epoch": 0.9065811288113602,
      "grad_norm": 0.14094327963979142,
      "learning_rate": 4.542885709148636e-06,
      "loss": 1.0061,
      "step": 12577
    },
    {
      "epoch": 0.9066532112736971,
      "grad_norm": 0.13359082573749986,
      "learning_rate": 4.535931254175463e-06,
      "loss": 1.0617,
      "step": 12578
    },
    {
      "epoch": 0.906725293736034,
      "grad_norm": 0.1295855560148549,
      "learning_rate": 4.5289820028480815e-06,
      "loss": 1.0807,
      "step": 12579
    },
    {
      "epoch": 0.906797376198371,
      "grad_norm": 0.12288911517626738,
      "learning_rate": 4.522037955545277e-06,
      "loss": 1.0105,
      "step": 12580
    },
    {
      "epoch": 0.9068694586607079,
      "grad_norm": 0.1136158428467369,
      "learning_rate": 4.515099112645571e-06,
      "loss": 1.0551,
      "step": 12581
    },
    {
      "epoch": 0.9069415411230448,
      "grad_norm": 0.11547050785061888,
      "learning_rate": 4.5081654745272266e-06,
      "loss": 1.0497,
      "step": 12582
    },
    {
      "epoch": 0.9070136235853816,
      "grad_norm": 0.11121748000801619,
      "learning_rate": 4.501237041568129e-06,
      "loss": 1.046,
      "step": 12583
    },
    {
      "epoch": 0.9070857060477185,
      "grad_norm": 0.18484608967905794,
      "learning_rate": 4.494313814145978e-06,
      "loss": 0.9818,
      "step": 12584
    },
    {
      "epoch": 0.9071577885100555,
      "grad_norm": 0.12842712893007782,
      "learning_rate": 4.487395792638149e-06,
      "loss": 1.1726,
      "step": 12585
    },
    {
      "epoch": 0.9072298709723924,
      "grad_norm": 0.2775820214333367,
      "learning_rate": 4.480482977421718e-06,
      "loss": 1.0388,
      "step": 12586
    },
    {
      "epoch": 0.9073019534347293,
      "grad_norm": 1.1468705602375244,
      "learning_rate": 4.473575368873517e-06,
      "loss": 1.0292,
      "step": 12587
    },
    {
      "epoch": 0.9073740358970662,
      "grad_norm": 0.13909261563204267,
      "learning_rate": 4.466672967370056e-06,
      "loss": 1.1006,
      "step": 12588
    },
    {
      "epoch": 0.9074461183594031,
      "grad_norm": 0.12557913711032084,
      "learning_rate": 4.459775773287578e-06,
      "loss": 1.0017,
      "step": 12589
    },
    {
      "epoch": 0.9075182008217401,
      "grad_norm": 0.1303468007045536,
      "learning_rate": 4.45288378700206e-06,
      "loss": 1.0315,
      "step": 12590
    },
    {
      "epoch": 0.907590283284077,
      "grad_norm": 0.12371366595485163,
      "learning_rate": 4.445997008889158e-06,
      "loss": 1.0745,
      "step": 12591
    },
    {
      "epoch": 0.9076623657464139,
      "grad_norm": 0.11256019630106179,
      "learning_rate": 4.439115439324271e-06,
      "loss": 1.006,
      "step": 12592
    },
    {
      "epoch": 0.9077344482087508,
      "grad_norm": 1.670190931907813,
      "learning_rate": 4.432239078682499e-06,
      "loss": 1.0037,
      "step": 12593
    },
    {
      "epoch": 0.9078065306710877,
      "grad_norm": 0.2031265724890723,
      "learning_rate": 4.425367927338675e-06,
      "loss": 1.0321,
      "step": 12594
    },
    {
      "epoch": 0.9078786131334247,
      "grad_norm": 0.11312145624594348,
      "learning_rate": 4.418501985667334e-06,
      "loss": 0.9055,
      "step": 12595
    },
    {
      "epoch": 0.9079506955957616,
      "grad_norm": 0.13603443648414,
      "learning_rate": 4.411641254042731e-06,
      "loss": 1.1181,
      "step": 12596
    },
    {
      "epoch": 0.9080227780580985,
      "grad_norm": 0.11299611901759313,
      "learning_rate": 4.404785732838846e-06,
      "loss": 1.1242,
      "step": 12597
    },
    {
      "epoch": 0.9080948605204354,
      "grad_norm": 0.1310578665126778,
      "learning_rate": 4.397935422429345e-06,
      "loss": 1.044,
      "step": 12598
    },
    {
      "epoch": 0.9081669429827723,
      "grad_norm": 0.19924209764373227,
      "learning_rate": 4.3910903231876524e-06,
      "loss": 1.0393,
      "step": 12599
    },
    {
      "epoch": 0.9082390254451093,
      "grad_norm": 0.14395792409877448,
      "learning_rate": 4.38425043548687e-06,
      "loss": 1.0469,
      "step": 12600
    },
    {
      "epoch": 0.9083111079074461,
      "grad_norm": 0.15888050339067883,
      "learning_rate": 4.377415759699854e-06,
      "loss": 1.0568,
      "step": 12601
    },
    {
      "epoch": 0.908383190369783,
      "grad_norm": 0.12317967343522196,
      "learning_rate": 4.37058629619913e-06,
      "loss": 1.0344,
      "step": 12602
    },
    {
      "epoch": 0.9084552728321199,
      "grad_norm": 0.14075749791992245,
      "learning_rate": 4.3637620453569896e-06,
      "loss": 1.0672,
      "step": 12603
    },
    {
      "epoch": 0.9085273552944568,
      "grad_norm": 0.11876130901521542,
      "learning_rate": 4.3569430075454e-06,
      "loss": 1.0208,
      "step": 12604
    },
    {
      "epoch": 0.9085994377567937,
      "grad_norm": 0.11278310156457375,
      "learning_rate": 4.3501291831360646e-06,
      "loss": 1.0491,
      "step": 12605
    },
    {
      "epoch": 0.9086715202191307,
      "grad_norm": 0.11225571310361603,
      "learning_rate": 4.3433205725003975e-06,
      "loss": 1.0282,
      "step": 12606
    },
    {
      "epoch": 0.9087436026814676,
      "grad_norm": 0.10990225973111731,
      "learning_rate": 4.336517176009535e-06,
      "loss": 0.965,
      "step": 12607
    },
    {
      "epoch": 0.9088156851438045,
      "grad_norm": 0.1281898350849102,
      "learning_rate": 4.329718994034293e-06,
      "loss": 1.017,
      "step": 12608
    },
    {
      "epoch": 0.9088877676061414,
      "grad_norm": 0.15597855804191468,
      "learning_rate": 4.3229260269452734e-06,
      "loss": 1.0852,
      "step": 12609
    },
    {
      "epoch": 0.9089598500684783,
      "grad_norm": 0.19744209531485313,
      "learning_rate": 4.316138275112735e-06,
      "loss": 0.9714,
      "step": 12610
    },
    {
      "epoch": 0.9090319325308153,
      "grad_norm": 0.11001572597240324,
      "learning_rate": 4.309355738906684e-06,
      "loss": 1.0721,
      "step": 12611
    },
    {
      "epoch": 0.9091040149931522,
      "grad_norm": 0.2830988859093341,
      "learning_rate": 4.30257841869679e-06,
      "loss": 0.9933,
      "step": 12612
    },
    {
      "epoch": 0.9091760974554891,
      "grad_norm": 0.11986480042174198,
      "learning_rate": 4.295806314852524e-06,
      "loss": 1.0399,
      "step": 12613
    },
    {
      "epoch": 0.909248179917826,
      "grad_norm": 0.14990653706431528,
      "learning_rate": 4.289039427743014e-06,
      "loss": 1.0583,
      "step": 12614
    },
    {
      "epoch": 0.909320262380163,
      "grad_norm": 0.11857193621889843,
      "learning_rate": 4.282277757737108e-06,
      "loss": 1.0532,
      "step": 12615
    },
    {
      "epoch": 0.9093923448424999,
      "grad_norm": 0.13185863797862893,
      "learning_rate": 4.275521305203367e-06,
      "loss": 1.0236,
      "step": 12616
    },
    {
      "epoch": 0.9094644273048368,
      "grad_norm": 0.1355535896036068,
      "learning_rate": 4.268770070510097e-06,
      "loss": 1.0371,
      "step": 12617
    },
    {
      "epoch": 0.9095365097671736,
      "grad_norm": 0.10469410867787449,
      "learning_rate": 4.2620240540252935e-06,
      "loss": 0.9855,
      "step": 12618
    },
    {
      "epoch": 0.9096085922295105,
      "grad_norm": 0.16039687845260836,
      "learning_rate": 4.255283256116682e-06,
      "loss": 1.0746,
      "step": 12619
    },
    {
      "epoch": 0.9096806746918474,
      "grad_norm": 0.13198063265325863,
      "learning_rate": 4.248547677151693e-06,
      "loss": 1.0457,
      "step": 12620
    },
    {
      "epoch": 0.9097527571541844,
      "grad_norm": 0.12588081332899437,
      "learning_rate": 4.241817317497443e-06,
      "loss": 1.045,
      "step": 12621
    },
    {
      "epoch": 0.9098248396165213,
      "grad_norm": 0.15159533299536418,
      "learning_rate": 4.235092177520849e-06,
      "loss": 1.0617,
      "step": 12622
    },
    {
      "epoch": 0.9098969220788582,
      "grad_norm": 0.12114872830937118,
      "learning_rate": 4.228372257588487e-06,
      "loss": 1.0359,
      "step": 12623
    },
    {
      "epoch": 0.9099690045411951,
      "grad_norm": 0.11223580974975442,
      "learning_rate": 4.221657558066605e-06,
      "loss": 1.0385,
      "step": 12624
    },
    {
      "epoch": 0.910041087003532,
      "grad_norm": 0.1298446396431837,
      "learning_rate": 4.214948079321246e-06,
      "loss": 0.9789,
      "step": 12625
    },
    {
      "epoch": 0.910113169465869,
      "grad_norm": 0.29448493530121267,
      "learning_rate": 4.2082438217181385e-06,
      "loss": 1.0552,
      "step": 12626
    },
    {
      "epoch": 0.9101852519282059,
      "grad_norm": 0.11204482782460502,
      "learning_rate": 4.201544785622735e-06,
      "loss": 1.045,
      "step": 12627
    },
    {
      "epoch": 0.9102573343905428,
      "grad_norm": 0.11916707288031698,
      "learning_rate": 4.194850971400166e-06,
      "loss": 0.9854,
      "step": 12628
    },
    {
      "epoch": 0.9103294168528797,
      "grad_norm": 0.15680009391916153,
      "learning_rate": 4.188162379415317e-06,
      "loss": 1.0301,
      "step": 12629
    },
    {
      "epoch": 0.9104014993152166,
      "grad_norm": 0.1328601592886331,
      "learning_rate": 4.181479010032763e-06,
      "loss": 1.0487,
      "step": 12630
    },
    {
      "epoch": 0.9104735817775536,
      "grad_norm": 0.11992029441437253,
      "learning_rate": 4.1748008636168455e-06,
      "loss": 0.9818,
      "step": 12631
    },
    {
      "epoch": 0.9105456642398905,
      "grad_norm": 0.12135570308443677,
      "learning_rate": 4.168127940531541e-06,
      "loss": 1.0408,
      "step": 12632
    },
    {
      "epoch": 0.9106177467022274,
      "grad_norm": 0.16582667819735922,
      "learning_rate": 4.161460241140602e-06,
      "loss": 1.036,
      "step": 12633
    },
    {
      "epoch": 0.9106898291645643,
      "grad_norm": 0.13416517593706515,
      "learning_rate": 4.154797765807472e-06,
      "loss": 1.0765,
      "step": 12634
    },
    {
      "epoch": 0.9107619116269011,
      "grad_norm": 0.11428059287558898,
      "learning_rate": 4.148140514895338e-06,
      "loss": 1.0371,
      "step": 12635
    },
    {
      "epoch": 0.910833994089238,
      "grad_norm": 0.12094363590926648,
      "learning_rate": 4.141488488767042e-06,
      "loss": 1.027,
      "step": 12636
    },
    {
      "epoch": 0.910906076551575,
      "grad_norm": 0.1323250972767189,
      "learning_rate": 4.134841687785207e-06,
      "loss": 1.0465,
      "step": 12637
    },
    {
      "epoch": 0.9109781590139119,
      "grad_norm": 0.13987526115113744,
      "learning_rate": 4.128200112312119e-06,
      "loss": 1.0137,
      "step": 12638
    },
    {
      "epoch": 0.9110502414762488,
      "grad_norm": 0.13856882777892296,
      "learning_rate": 4.121563762709846e-06,
      "loss": 1.0069,
      "step": 12639
    },
    {
      "epoch": 0.9111223239385857,
      "grad_norm": 0.13262682712684556,
      "learning_rate": 4.114932639340074e-06,
      "loss": 0.9809,
      "step": 12640
    },
    {
      "epoch": 0.9111944064009226,
      "grad_norm": 0.11612085292936113,
      "learning_rate": 4.1083067425643054e-06,
      "loss": 1.0665,
      "step": 12641
    },
    {
      "epoch": 0.9112664888632596,
      "grad_norm": 0.1297219441331253,
      "learning_rate": 4.1016860727436826e-06,
      "loss": 1.0457,
      "step": 12642
    },
    {
      "epoch": 0.9113385713255965,
      "grad_norm": 0.110701927666298,
      "learning_rate": 4.095070630239095e-06,
      "loss": 1.1335,
      "step": 12643
    },
    {
      "epoch": 0.9114106537879334,
      "grad_norm": 0.11210756863097575,
      "learning_rate": 4.088460415411155e-06,
      "loss": 1.0952,
      "step": 12644
    },
    {
      "epoch": 0.9114827362502703,
      "grad_norm": 0.11753727291294688,
      "learning_rate": 4.081855428620174e-06,
      "loss": 1.0447,
      "step": 12645
    },
    {
      "epoch": 0.9115548187126072,
      "grad_norm": 0.11746411596811551,
      "learning_rate": 4.075255670226175e-06,
      "loss": 1.0389,
      "step": 12646
    },
    {
      "epoch": 0.9116269011749442,
      "grad_norm": 0.11429899421596,
      "learning_rate": 4.068661140588903e-06,
      "loss": 1.008,
      "step": 12647
    },
    {
      "epoch": 0.9116989836372811,
      "grad_norm": 0.1297615657402309,
      "learning_rate": 4.062071840067838e-06,
      "loss": 1.0139,
      "step": 12648
    },
    {
      "epoch": 0.911771066099618,
      "grad_norm": 0.1307294855761369,
      "learning_rate": 4.055487769022137e-06,
      "loss": 1.0168,
      "step": 12649
    },
    {
      "epoch": 0.9118431485619549,
      "grad_norm": 0.16020706409643481,
      "learning_rate": 4.04890892781069e-06,
      "loss": 0.9946,
      "step": 12650
    },
    {
      "epoch": 0.9119152310242918,
      "grad_norm": 0.15596711907945535,
      "learning_rate": 4.042335316792123e-06,
      "loss": 1.0019,
      "step": 12651
    },
    {
      "epoch": 0.9119873134866286,
      "grad_norm": 0.11102970375595746,
      "learning_rate": 4.035766936324736e-06,
      "loss": 0.9842,
      "step": 12652
    },
    {
      "epoch": 0.9120593959489656,
      "grad_norm": 0.12435902743347975,
      "learning_rate": 4.029203786766567e-06,
      "loss": 1.0105,
      "step": 12653
    },
    {
      "epoch": 0.9121314784113025,
      "grad_norm": 0.10751422508111999,
      "learning_rate": 4.022645868475372e-06,
      "loss": 1.0967,
      "step": 12654
    },
    {
      "epoch": 0.9122035608736394,
      "grad_norm": 0.11979907925689307,
      "learning_rate": 4.016093181808623e-06,
      "loss": 0.983,
      "step": 12655
    },
    {
      "epoch": 0.9122756433359763,
      "grad_norm": 0.13619716685029373,
      "learning_rate": 4.009545727123487e-06,
      "loss": 1.0788,
      "step": 12656
    },
    {
      "epoch": 0.9123477257983132,
      "grad_norm": 0.1275404521227242,
      "learning_rate": 4.00300350477687e-06,
      "loss": 1.1151,
      "step": 12657
    },
    {
      "epoch": 0.9124198082606502,
      "grad_norm": 0.11607962666365793,
      "learning_rate": 3.996466515125363e-06,
      "loss": 0.9942,
      "step": 12658
    },
    {
      "epoch": 0.9124918907229871,
      "grad_norm": 0.12256889429959665,
      "learning_rate": 3.989934758525315e-06,
      "loss": 1.0592,
      "step": 12659
    },
    {
      "epoch": 0.912563973185324,
      "grad_norm": 0.12148988326441822,
      "learning_rate": 3.983408235332753e-06,
      "loss": 1.0042,
      "step": 12660
    },
    {
      "epoch": 0.9126360556476609,
      "grad_norm": 0.1325188714720689,
      "learning_rate": 3.9768869459034245e-06,
      "loss": 1.001,
      "step": 12661
    },
    {
      "epoch": 0.9127081381099978,
      "grad_norm": 0.1243715440906763,
      "learning_rate": 3.9703708905928005e-06,
      "loss": 0.9695,
      "step": 12662
    },
    {
      "epoch": 0.9127802205723348,
      "grad_norm": 0.13686890983304142,
      "learning_rate": 3.963860069756075e-06,
      "loss": 1.0161,
      "step": 12663
    },
    {
      "epoch": 0.9128523030346717,
      "grad_norm": 0.11762684269699401,
      "learning_rate": 3.957354483748154e-06,
      "loss": 1.0157,
      "step": 12664
    },
    {
      "epoch": 0.9129243854970086,
      "grad_norm": 0.13526449412467934,
      "learning_rate": 3.95085413292361e-06,
      "loss": 1.1031,
      "step": 12665
    },
    {
      "epoch": 0.9129964679593455,
      "grad_norm": 0.11856500528618782,
      "learning_rate": 3.944359017636812e-06,
      "loss": 1.0368,
      "step": 12666
    },
    {
      "epoch": 0.9130685504216824,
      "grad_norm": 0.12295281408575191,
      "learning_rate": 3.9378691382417784e-06,
      "loss": 1.0401,
      "step": 12667
    },
    {
      "epoch": 0.9131406328840194,
      "grad_norm": 0.38315908465933,
      "learning_rate": 3.931384495092283e-06,
      "loss": 1.0889,
      "step": 12668
    },
    {
      "epoch": 0.9132127153463563,
      "grad_norm": 0.1316676148531167,
      "learning_rate": 3.924905088541775e-06,
      "loss": 1.0461,
      "step": 12669
    },
    {
      "epoch": 0.9132847978086931,
      "grad_norm": 0.11820229207334432,
      "learning_rate": 3.91843091894345e-06,
      "loss": 1.0388,
      "step": 12670
    },
    {
      "epoch": 0.91335688027103,
      "grad_norm": 0.12521792818979582,
      "learning_rate": 3.911961986650226e-06,
      "loss": 1.0879,
      "step": 12671
    },
    {
      "epoch": 0.9134289627333669,
      "grad_norm": 0.13862378361339242,
      "learning_rate": 3.905498292014698e-06,
      "loss": 1.0259,
      "step": 12672
    },
    {
      "epoch": 0.9135010451957039,
      "grad_norm": 0.11329774932473564,
      "learning_rate": 3.899039835389196e-06,
      "loss": 1.0655,
      "step": 12673
    },
    {
      "epoch": 0.9135731276580408,
      "grad_norm": 0.12418672758458654,
      "learning_rate": 3.89258661712576e-06,
      "loss": 1.0657,
      "step": 12674
    },
    {
      "epoch": 0.9136452101203777,
      "grad_norm": 0.12674562434348632,
      "learning_rate": 3.886138637576153e-06,
      "loss": 1.0567,
      "step": 12675
    },
    {
      "epoch": 0.9137172925827146,
      "grad_norm": 0.1386320777403062,
      "learning_rate": 3.879695897091873e-06,
      "loss": 1.0118,
      "step": 12676
    },
    {
      "epoch": 0.9137893750450515,
      "grad_norm": 0.12305515362486003,
      "learning_rate": 3.87325839602406e-06,
      "loss": 1.0608,
      "step": 12677
    },
    {
      "epoch": 0.9138614575073885,
      "grad_norm": 0.1558497812639037,
      "learning_rate": 3.866826134723633e-06,
      "loss": 0.9561,
      "step": 12678
    },
    {
      "epoch": 0.9139335399697254,
      "grad_norm": 0.5399128452625257,
      "learning_rate": 3.860399113541235e-06,
      "loss": 0.9605,
      "step": 12679
    },
    {
      "epoch": 0.9140056224320623,
      "grad_norm": 0.11461385289279087,
      "learning_rate": 3.8539773328271745e-06,
      "loss": 1.0454,
      "step": 12680
    },
    {
      "epoch": 0.9140777048943992,
      "grad_norm": 0.11726514374535914,
      "learning_rate": 3.847560792931482e-06,
      "loss": 1.1034,
      "step": 12681
    },
    {
      "epoch": 0.9141497873567361,
      "grad_norm": 0.11389184406676857,
      "learning_rate": 3.841149494203933e-06,
      "loss": 0.9767,
      "step": 12682
    },
    {
      "epoch": 0.914221869819073,
      "grad_norm": 1.781297494051313,
      "learning_rate": 3.834743436993993e-06,
      "loss": 0.9987,
      "step": 12683
    },
    {
      "epoch": 0.91429395228141,
      "grad_norm": 0.13363605641935697,
      "learning_rate": 3.828342621650882e-06,
      "loss": 1.0575,
      "step": 12684
    },
    {
      "epoch": 0.9143660347437469,
      "grad_norm": 0.14238346281937597,
      "learning_rate": 3.8219470485234555e-06,
      "loss": 1.0942,
      "step": 12685
    },
    {
      "epoch": 0.9144381172060838,
      "grad_norm": 0.11864959598836702,
      "learning_rate": 3.815556717960345e-06,
      "loss": 1.0315,
      "step": 12686
    },
    {
      "epoch": 0.9145101996684206,
      "grad_norm": 0.13492316575323138,
      "learning_rate": 3.8091716303098714e-06,
      "loss": 1.0047,
      "step": 12687
    },
    {
      "epoch": 0.9145822821307575,
      "grad_norm": 0.1291411377410279,
      "learning_rate": 3.8027917859201234e-06,
      "loss": 1.0419,
      "step": 12688
    },
    {
      "epoch": 0.9146543645930945,
      "grad_norm": 0.13052181058206477,
      "learning_rate": 3.796417185138812e-06,
      "loss": 1.0656,
      "step": 12689
    },
    {
      "epoch": 0.9147264470554314,
      "grad_norm": 0.11598633877239947,
      "learning_rate": 3.7900478283134255e-06,
      "loss": 1.0076,
      "step": 12690
    },
    {
      "epoch": 0.9147985295177683,
      "grad_norm": 0.12935078402354055,
      "learning_rate": 3.783683715791142e-06,
      "loss": 1.0772,
      "step": 12691
    },
    {
      "epoch": 0.9148706119801052,
      "grad_norm": 0.1434897402437985,
      "learning_rate": 3.777324847918895e-06,
      "loss": 1.0587,
      "step": 12692
    },
    {
      "epoch": 0.9149426944424421,
      "grad_norm": 0.14337790133042855,
      "learning_rate": 3.7709712250432627e-06,
      "loss": 0.909,
      "step": 12693
    },
    {
      "epoch": 0.9150147769047791,
      "grad_norm": 0.19135638115518988,
      "learning_rate": 3.7646228475105905e-06,
      "loss": 1.0434,
      "step": 12694
    },
    {
      "epoch": 0.915086859367116,
      "grad_norm": 0.14561177727400196,
      "learning_rate": 3.7582797156669125e-06,
      "loss": 0.9653,
      "step": 12695
    },
    {
      "epoch": 0.9151589418294529,
      "grad_norm": 0.11117580045694338,
      "learning_rate": 3.7519418298579966e-06,
      "loss": 1.0655,
      "step": 12696
    },
    {
      "epoch": 0.9152310242917898,
      "grad_norm": 0.12989913895843888,
      "learning_rate": 3.7456091904293113e-06,
      "loss": 1.112,
      "step": 12697
    },
    {
      "epoch": 0.9153031067541267,
      "grad_norm": 0.12542930975308017,
      "learning_rate": 3.7392817977260464e-06,
      "loss": 1.0752,
      "step": 12698
    },
    {
      "epoch": 0.9153751892164637,
      "grad_norm": 0.13592914563073463,
      "learning_rate": 3.732959652093093e-06,
      "loss": 0.996,
      "step": 12699
    },
    {
      "epoch": 0.9154472716788006,
      "grad_norm": 0.13095764110885874,
      "learning_rate": 3.7266427538750647e-06,
      "loss": 1.0361,
      "step": 12700
    },
    {
      "epoch": 0.9155193541411375,
      "grad_norm": 0.1082965602870162,
      "learning_rate": 3.720331103416297e-06,
      "loss": 1.0139,
      "step": 12701
    },
    {
      "epoch": 0.9155914366034744,
      "grad_norm": 0.11914659638598349,
      "learning_rate": 3.7140247010608255e-06,
      "loss": 1.011,
      "step": 12702
    },
    {
      "epoch": 0.9156635190658113,
      "grad_norm": 0.1209059199541609,
      "learning_rate": 3.7077235471523974e-06,
      "loss": 1.1043,
      "step": 12703
    },
    {
      "epoch": 0.9157356015281481,
      "grad_norm": 0.11319476180727449,
      "learning_rate": 3.701427642034505e-06,
      "loss": 1.1512,
      "step": 12704
    },
    {
      "epoch": 0.9158076839904851,
      "grad_norm": 0.12228777424112541,
      "learning_rate": 3.695136986050307e-06,
      "loss": 1.0143,
      "step": 12705
    },
    {
      "epoch": 0.915879766452822,
      "grad_norm": 0.2920570400923903,
      "learning_rate": 3.688851579542718e-06,
      "loss": 1.022,
      "step": 12706
    },
    {
      "epoch": 0.9159518489151589,
      "grad_norm": 0.1310655692902544,
      "learning_rate": 3.6825714228543416e-06,
      "loss": 1.0166,
      "step": 12707
    },
    {
      "epoch": 0.9160239313774958,
      "grad_norm": 0.1286693251631029,
      "learning_rate": 3.6762965163275044e-06,
      "loss": 0.9938,
      "step": 12708
    },
    {
      "epoch": 0.9160960138398327,
      "grad_norm": 0.14695090249984824,
      "learning_rate": 3.6700268603042433e-06,
      "loss": 1.0642,
      "step": 12709
    },
    {
      "epoch": 0.9161680963021697,
      "grad_norm": 0.13867288855141707,
      "learning_rate": 3.6637624551263073e-06,
      "loss": 1.0678,
      "step": 12710
    },
    {
      "epoch": 0.9162401787645066,
      "grad_norm": 0.12218168378189115,
      "learning_rate": 3.657503301135168e-06,
      "loss": 1.05,
      "step": 12711
    },
    {
      "epoch": 0.9163122612268435,
      "grad_norm": 0.1454607631764603,
      "learning_rate": 3.6512493986719963e-06,
      "loss": 1.0273,
      "step": 12712
    },
    {
      "epoch": 0.9163843436891804,
      "grad_norm": 0.1415171884958042,
      "learning_rate": 3.6450007480777093e-06,
      "loss": 1.07,
      "step": 12713
    },
    {
      "epoch": 0.9164564261515173,
      "grad_norm": 0.10971092045872456,
      "learning_rate": 3.6387573496928784e-06,
      "loss": 1.0403,
      "step": 12714
    },
    {
      "epoch": 0.9165285086138543,
      "grad_norm": 0.12432907083235464,
      "learning_rate": 3.6325192038578536e-06,
      "loss": 1.0139,
      "step": 12715
    },
    {
      "epoch": 0.9166005910761912,
      "grad_norm": 0.12905133336075594,
      "learning_rate": 3.626286310912663e-06,
      "loss": 1.1142,
      "step": 12716
    },
    {
      "epoch": 0.9166726735385281,
      "grad_norm": 0.11715361177647896,
      "learning_rate": 3.6200586711970685e-06,
      "loss": 1.045,
      "step": 12717
    },
    {
      "epoch": 0.916744756000865,
      "grad_norm": 0.13231251653911855,
      "learning_rate": 3.613836285050487e-06,
      "loss": 0.9243,
      "step": 12718
    },
    {
      "epoch": 0.9168168384632019,
      "grad_norm": 0.10464899338950145,
      "learning_rate": 3.607619152812125e-06,
      "loss": 0.9628,
      "step": 12719
    },
    {
      "epoch": 0.9168889209255389,
      "grad_norm": 0.1228696092280143,
      "learning_rate": 3.601407274820878e-06,
      "loss": 0.9799,
      "step": 12720
    },
    {
      "epoch": 0.9169610033878758,
      "grad_norm": 0.15169782912323898,
      "learning_rate": 3.5952006514153645e-06,
      "loss": 0.9879,
      "step": 12721
    },
    {
      "epoch": 0.9170330858502126,
      "grad_norm": 0.12904020711184808,
      "learning_rate": 3.588999282933836e-06,
      "loss": 1.0562,
      "step": 12722
    },
    {
      "epoch": 0.9171051683125495,
      "grad_norm": 0.11503676231343958,
      "learning_rate": 3.582803169714388e-06,
      "loss": 1.0608,
      "step": 12723
    },
    {
      "epoch": 0.9171772507748864,
      "grad_norm": 0.1276551764505408,
      "learning_rate": 3.576612312094729e-06,
      "loss": 1.0766,
      "step": 12724
    },
    {
      "epoch": 0.9172493332372234,
      "grad_norm": 0.10627084065147567,
      "learning_rate": 3.5704267104123444e-06,
      "loss": 1.0107,
      "step": 12725
    },
    {
      "epoch": 0.9173214156995603,
      "grad_norm": 0.12513942913886733,
      "learning_rate": 3.5642463650043644e-06,
      "loss": 1.0565,
      "step": 12726
    },
    {
      "epoch": 0.9173934981618972,
      "grad_norm": 0.20870043347692807,
      "learning_rate": 3.5580712762077084e-06,
      "loss": 1.1098,
      "step": 12727
    },
    {
      "epoch": 0.9174655806242341,
      "grad_norm": 0.14857986979373805,
      "learning_rate": 3.551901444358974e-06,
      "loss": 0.9957,
      "step": 12728
    },
    {
      "epoch": 0.917537663086571,
      "grad_norm": 0.11867857430919863,
      "learning_rate": 3.545736869794458e-06,
      "loss": 1.039,
      "step": 12729
    },
    {
      "epoch": 0.917609745548908,
      "grad_norm": 0.1265182827341246,
      "learning_rate": 3.5395775528501706e-06,
      "loss": 1.0431,
      "step": 12730
    },
    {
      "epoch": 0.9176818280112449,
      "grad_norm": 0.1181897435548992,
      "learning_rate": 3.533423493861865e-06,
      "loss": 1.0268,
      "step": 12731
    },
    {
      "epoch": 0.9177539104735818,
      "grad_norm": 0.13300483074417135,
      "learning_rate": 3.527274693165017e-06,
      "loss": 1.0497,
      "step": 12732
    },
    {
      "epoch": 0.9178259929359187,
      "grad_norm": 0.1259485067756669,
      "learning_rate": 3.5211311510947475e-06,
      "loss": 1.0412,
      "step": 12733
    },
    {
      "epoch": 0.9178980753982556,
      "grad_norm": 0.11781499416106103,
      "learning_rate": 3.5149928679859555e-06,
      "loss": 0.9839,
      "step": 12734
    },
    {
      "epoch": 0.9179701578605926,
      "grad_norm": 0.12281929741027424,
      "learning_rate": 3.508859844173229e-06,
      "loss": 1.0083,
      "step": 12735
    },
    {
      "epoch": 0.9180422403229295,
      "grad_norm": 0.11901592578013359,
      "learning_rate": 3.5027320799909e-06,
      "loss": 1.0055,
      "step": 12736
    },
    {
      "epoch": 0.9181143227852664,
      "grad_norm": 0.1533347778534519,
      "learning_rate": 3.4966095757729356e-06,
      "loss": 1.0781,
      "step": 12737
    },
    {
      "epoch": 0.9181864052476033,
      "grad_norm": 0.12570877875281797,
      "learning_rate": 3.490492331853101e-06,
      "loss": 0.9572,
      "step": 12738
    },
    {
      "epoch": 0.9182584877099401,
      "grad_norm": 0.12029825424349053,
      "learning_rate": 3.4843803485648417e-06,
      "loss": 0.9837,
      "step": 12739
    },
    {
      "epoch": 0.918330570172277,
      "grad_norm": 0.12352664497205941,
      "learning_rate": 3.478273626241302e-06,
      "loss": 1.0042,
      "step": 12740
    },
    {
      "epoch": 0.918402652634614,
      "grad_norm": 0.12481596178028524,
      "learning_rate": 3.472172165215348e-06,
      "loss": 1.0637,
      "step": 12741
    },
    {
      "epoch": 0.9184747350969509,
      "grad_norm": 0.123460331828719,
      "learning_rate": 3.466075965819582e-06,
      "loss": 0.9019,
      "step": 12742
    },
    {
      "epoch": 0.9185468175592878,
      "grad_norm": 0.12064789816037653,
      "learning_rate": 3.459985028386292e-06,
      "loss": 1.017,
      "step": 12743
    },
    {
      "epoch": 0.9186189000216247,
      "grad_norm": 0.12769195128837588,
      "learning_rate": 3.4538993532474918e-06,
      "loss": 1.025,
      "step": 12744
    },
    {
      "epoch": 0.9186909824839616,
      "grad_norm": 0.15803273389779832,
      "learning_rate": 3.447818940734904e-06,
      "loss": 1.0353,
      "step": 12745
    },
    {
      "epoch": 0.9187630649462986,
      "grad_norm": 0.11513321742495279,
      "learning_rate": 3.4417437911799523e-06,
      "loss": 1.0264,
      "step": 12746
    },
    {
      "epoch": 0.9188351474086355,
      "grad_norm": 0.10982059219397958,
      "learning_rate": 3.4356739049138166e-06,
      "loss": 1.0049,
      "step": 12747
    },
    {
      "epoch": 0.9189072298709724,
      "grad_norm": 0.13533489567666993,
      "learning_rate": 3.4296092822673322e-06,
      "loss": 1.0317,
      "step": 12748
    },
    {
      "epoch": 0.9189793123333093,
      "grad_norm": 0.1523917353478114,
      "learning_rate": 3.4235499235710788e-06,
      "loss": 1.0488,
      "step": 12749
    },
    {
      "epoch": 0.9190513947956462,
      "grad_norm": 0.11345874689479306,
      "learning_rate": 3.4174958291553593e-06,
      "loss": 0.9632,
      "step": 12750
    },
    {
      "epoch": 0.9191234772579832,
      "grad_norm": 0.15947417205238368,
      "learning_rate": 3.411446999350176e-06,
      "loss": 1.0231,
      "step": 12751
    },
    {
      "epoch": 0.9191955597203201,
      "grad_norm": 0.13033704221423134,
      "learning_rate": 3.4054034344852326e-06,
      "loss": 1.0508,
      "step": 12752
    },
    {
      "epoch": 0.919267642182657,
      "grad_norm": 0.11673329878582411,
      "learning_rate": 3.3993651348899538e-06,
      "loss": 1.0621,
      "step": 12753
    },
    {
      "epoch": 0.9193397246449939,
      "grad_norm": 0.11501594850625887,
      "learning_rate": 3.3933321008934984e-06,
      "loss": 1.0148,
      "step": 12754
    },
    {
      "epoch": 0.9194118071073308,
      "grad_norm": 0.12467056924683186,
      "learning_rate": 3.3873043328247032e-06,
      "loss": 1.0393,
      "step": 12755
    },
    {
      "epoch": 0.9194838895696676,
      "grad_norm": 0.10592818280545116,
      "learning_rate": 3.3812818310121507e-06,
      "loss": 0.9934,
      "step": 12756
    },
    {
      "epoch": 0.9195559720320046,
      "grad_norm": 0.11088084517260449,
      "learning_rate": 3.3752645957841112e-06,
      "loss": 0.9395,
      "step": 12757
    },
    {
      "epoch": 0.9196280544943415,
      "grad_norm": 0.12427955615736506,
      "learning_rate": 3.3692526274685776e-06,
      "loss": 0.9949,
      "step": 12758
    },
    {
      "epoch": 0.9197001369566784,
      "grad_norm": 0.11155642251790912,
      "learning_rate": 3.3632459263932657e-06,
      "loss": 1.0039,
      "step": 12759
    },
    {
      "epoch": 0.9197722194190153,
      "grad_norm": 0.12029222885162366,
      "learning_rate": 3.357244492885592e-06,
      "loss": 0.9974,
      "step": 12760
    },
    {
      "epoch": 0.9198443018813522,
      "grad_norm": 0.13388259086196275,
      "learning_rate": 3.3512483272726826e-06,
      "loss": 1.0519,
      "step": 12761
    },
    {
      "epoch": 0.9199163843436892,
      "grad_norm": 0.12800021533504338,
      "learning_rate": 3.3452574298813653e-06,
      "loss": 1.0581,
      "step": 12762
    },
    {
      "epoch": 0.9199884668060261,
      "grad_norm": 0.23098951982002808,
      "learning_rate": 3.339271801038235e-06,
      "loss": 1.0854,
      "step": 12763
    },
    {
      "epoch": 0.920060549268363,
      "grad_norm": 0.15167342683831977,
      "learning_rate": 3.3332914410695415e-06,
      "loss": 1.0446,
      "step": 12764
    },
    {
      "epoch": 0.9201326317306999,
      "grad_norm": 0.17617389926762705,
      "learning_rate": 3.327316350301268e-06,
      "loss": 0.9947,
      "step": 12765
    },
    {
      "epoch": 0.9202047141930368,
      "grad_norm": 0.17091493272450048,
      "learning_rate": 3.3213465290590996e-06,
      "loss": 0.9846,
      "step": 12766
    },
    {
      "epoch": 0.9202767966553738,
      "grad_norm": 0.1401082927160493,
      "learning_rate": 3.315381977668475e-06,
      "loss": 1.0607,
      "step": 12767
    },
    {
      "epoch": 0.9203488791177107,
      "grad_norm": 0.14189964881405315,
      "learning_rate": 3.3094226964544893e-06,
      "loss": 1.0161,
      "step": 12768
    },
    {
      "epoch": 0.9204209615800476,
      "grad_norm": 0.133232359425025,
      "learning_rate": 3.303468685741995e-06,
      "loss": 1.0884,
      "step": 12769
    },
    {
      "epoch": 0.9204930440423845,
      "grad_norm": 0.10915508173768379,
      "learning_rate": 3.2975199458555207e-06,
      "loss": 1.0196,
      "step": 12770
    },
    {
      "epoch": 0.9205651265047214,
      "grad_norm": 0.1092288975058612,
      "learning_rate": 3.2915764771193292e-06,
      "loss": 1.0266,
      "step": 12771
    },
    {
      "epoch": 0.9206372089670584,
      "grad_norm": 0.11399387650657473,
      "learning_rate": 3.2856382798573947e-06,
      "loss": 1.0125,
      "step": 12772
    },
    {
      "epoch": 0.9207092914293953,
      "grad_norm": 0.1204350724405444,
      "learning_rate": 3.2797053543934255e-06,
      "loss": 0.9633,
      "step": 12773
    },
    {
      "epoch": 0.9207813738917321,
      "grad_norm": 0.13673146933295616,
      "learning_rate": 3.273777701050795e-06,
      "loss": 1.0236,
      "step": 12774
    },
    {
      "epoch": 0.920853456354069,
      "grad_norm": 0.1273905660832863,
      "learning_rate": 3.2678553201525907e-06,
      "loss": 1.0093,
      "step": 12775
    },
    {
      "epoch": 0.9209255388164059,
      "grad_norm": 0.11794370013034274,
      "learning_rate": 3.2619382120216868e-06,
      "loss": 0.9122,
      "step": 12776
    },
    {
      "epoch": 0.9209976212787429,
      "grad_norm": 0.1330103365172953,
      "learning_rate": 3.256026376980592e-06,
      "loss": 1.0475,
      "step": 12777
    },
    {
      "epoch": 0.9210697037410798,
      "grad_norm": 0.1351886766489396,
      "learning_rate": 3.2501198153515598e-06,
      "loss": 0.9889,
      "step": 12778
    },
    {
      "epoch": 0.9211417862034167,
      "grad_norm": 0.10751132383496713,
      "learning_rate": 3.2442185274565217e-06,
      "loss": 1.018,
      "step": 12779
    },
    {
      "epoch": 0.9212138686657536,
      "grad_norm": 0.13966460638145314,
      "learning_rate": 3.238322513617198e-06,
      "loss": 1.0686,
      "step": 12780
    },
    {
      "epoch": 0.9212859511280905,
      "grad_norm": 0.1324847154683806,
      "learning_rate": 3.2324317741549537e-06,
      "loss": 1.0613,
      "step": 12781
    },
    {
      "epoch": 0.9213580335904274,
      "grad_norm": 0.11349477647262497,
      "learning_rate": 3.2265463093908766e-06,
      "loss": 1.0471,
      "step": 12782
    },
    {
      "epoch": 0.9214301160527644,
      "grad_norm": 0.12032033907883494,
      "learning_rate": 3.2206661196457876e-06,
      "loss": 1.0385,
      "step": 12783
    },
    {
      "epoch": 0.9215021985151013,
      "grad_norm": 0.12321371999624432,
      "learning_rate": 3.2147912052401972e-06,
      "loss": 1.0006,
      "step": 12784
    },
    {
      "epoch": 0.9215742809774382,
      "grad_norm": 0.12038990406183216,
      "learning_rate": 3.2089215664943716e-06,
      "loss": 1.0014,
      "step": 12785
    },
    {
      "epoch": 0.9216463634397751,
      "grad_norm": 0.12517912781341267,
      "learning_rate": 3.2030572037282325e-06,
      "loss": 1.036,
      "step": 12786
    },
    {
      "epoch": 0.921718445902112,
      "grad_norm": 0.1366421301208189,
      "learning_rate": 3.1971981172614464e-06,
      "loss": 0.9875,
      "step": 12787
    },
    {
      "epoch": 0.921790528364449,
      "grad_norm": 0.20294337099160306,
      "learning_rate": 3.1913443074133687e-06,
      "loss": 1.0775,
      "step": 12788
    },
    {
      "epoch": 0.9218626108267859,
      "grad_norm": 0.13000545650988096,
      "learning_rate": 3.1854957745031223e-06,
      "loss": 1.1194,
      "step": 12789
    },
    {
      "epoch": 0.9219346932891228,
      "grad_norm": 0.13010481056779083,
      "learning_rate": 3.1796525188494743e-06,
      "loss": 1.0453,
      "step": 12790
    },
    {
      "epoch": 0.9220067757514596,
      "grad_norm": 0.12796516263651625,
      "learning_rate": 3.1738145407709476e-06,
      "loss": 0.9705,
      "step": 12791
    },
    {
      "epoch": 0.9220788582137965,
      "grad_norm": 0.19546880992211366,
      "learning_rate": 3.1679818405857544e-06,
      "loss": 1.0,
      "step": 12792
    },
    {
      "epoch": 0.9221509406761335,
      "grad_norm": 0.11894522131491446,
      "learning_rate": 3.1621544186118403e-06,
      "loss": 1.0277,
      "step": 12793
    },
    {
      "epoch": 0.9222230231384704,
      "grad_norm": 0.14627741046850448,
      "learning_rate": 3.15633227516684e-06,
      "loss": 0.9886,
      "step": 12794
    },
    {
      "epoch": 0.9222951056008073,
      "grad_norm": 0.1318004112143706,
      "learning_rate": 3.1505154105681223e-06,
      "loss": 1.0682,
      "step": 12795
    },
    {
      "epoch": 0.9223671880631442,
      "grad_norm": 0.10897001066948382,
      "learning_rate": 3.1447038251327554e-06,
      "loss": 1.0244,
      "step": 12796
    },
    {
      "epoch": 0.9224392705254811,
      "grad_norm": 0.13465088819838802,
      "learning_rate": 3.1388975191775196e-06,
      "loss": 1.0864,
      "step": 12797
    },
    {
      "epoch": 0.9225113529878181,
      "grad_norm": 0.14129412118866894,
      "learning_rate": 3.1330964930189168e-06,
      "loss": 1.0209,
      "step": 12798
    },
    {
      "epoch": 0.922583435450155,
      "grad_norm": 0.11731480410567012,
      "learning_rate": 3.12730074697315e-06,
      "loss": 0.9981,
      "step": 12799
    },
    {
      "epoch": 0.9226555179124919,
      "grad_norm": 0.13461545677325196,
      "learning_rate": 3.1215102813561435e-06,
      "loss": 1.0697,
      "step": 12800
    },
    {
      "epoch": 0.9227276003748288,
      "grad_norm": 0.1334465458190896,
      "learning_rate": 3.1157250964835237e-06,
      "loss": 0.9967,
      "step": 12801
    },
    {
      "epoch": 0.9227996828371657,
      "grad_norm": 0.1231212820687193,
      "learning_rate": 3.109945192670638e-06,
      "loss": 1.0104,
      "step": 12802
    },
    {
      "epoch": 0.9228717652995027,
      "grad_norm": 0.1191816639044983,
      "learning_rate": 3.1041705702325343e-06,
      "loss": 1.053,
      "step": 12803
    },
    {
      "epoch": 0.9229438477618396,
      "grad_norm": 0.24304499292871173,
      "learning_rate": 3.0984012294839937e-06,
      "loss": 1.077,
      "step": 12804
    },
    {
      "epoch": 0.9230159302241765,
      "grad_norm": 0.15229052989271208,
      "learning_rate": 3.0926371707394985e-06,
      "loss": 1.083,
      "step": 12805
    },
    {
      "epoch": 0.9230880126865134,
      "grad_norm": 0.125440029207132,
      "learning_rate": 3.0868783943132305e-06,
      "loss": 1.0202,
      "step": 12806
    },
    {
      "epoch": 0.9231600951488503,
      "grad_norm": 0.12361247990539936,
      "learning_rate": 3.0811249005191056e-06,
      "loss": 1.0669,
      "step": 12807
    },
    {
      "epoch": 0.9232321776111871,
      "grad_norm": 0.11373762053406059,
      "learning_rate": 3.075376689670728e-06,
      "loss": 1.0916,
      "step": 12808
    },
    {
      "epoch": 0.9233042600735241,
      "grad_norm": 0.14028114729911012,
      "learning_rate": 3.0696337620814364e-06,
      "loss": 1.0493,
      "step": 12809
    },
    {
      "epoch": 0.923376342535861,
      "grad_norm": 0.10998376798910664,
      "learning_rate": 3.0638961180642687e-06,
      "loss": 1.0102,
      "step": 12810
    },
    {
      "epoch": 0.9234484249981979,
      "grad_norm": 0.131190455844296,
      "learning_rate": 3.0581637579319755e-06,
      "loss": 1.0824,
      "step": 12811
    },
    {
      "epoch": 0.9235205074605348,
      "grad_norm": 0.13280685397626238,
      "learning_rate": 3.0524366819970287e-06,
      "loss": 0.9965,
      "step": 12812
    },
    {
      "epoch": 0.9235925899228717,
      "grad_norm": 0.24685694308813688,
      "learning_rate": 3.0467148905716003e-06,
      "loss": 1.0565,
      "step": 12813
    },
    {
      "epoch": 0.9236646723852087,
      "grad_norm": 0.3519436563588596,
      "learning_rate": 3.040998383967586e-06,
      "loss": 1.0919,
      "step": 12814
    },
    {
      "epoch": 0.9237367548475456,
      "grad_norm": 0.13427325651078362,
      "learning_rate": 3.0352871624965584e-06,
      "loss": 1.1135,
      "step": 12815
    },
    {
      "epoch": 0.9238088373098825,
      "grad_norm": 0.14665952463777066,
      "learning_rate": 3.0295812264698687e-06,
      "loss": 1.0344,
      "step": 12816
    },
    {
      "epoch": 0.9238809197722194,
      "grad_norm": 0.16191972597341533,
      "learning_rate": 3.023880576198512e-06,
      "loss": 0.9848,
      "step": 12817
    },
    {
      "epoch": 0.9239530022345563,
      "grad_norm": 0.12787677385386878,
      "learning_rate": 3.018185211993252e-06,
      "loss": 1.038,
      "step": 12818
    },
    {
      "epoch": 0.9240250846968933,
      "grad_norm": 0.11700668886113397,
      "learning_rate": 3.012495134164506e-06,
      "loss": 0.9976,
      "step": 12819
    },
    {
      "epoch": 0.9240971671592302,
      "grad_norm": 0.13464176932033575,
      "learning_rate": 3.006810343022459e-06,
      "loss": 1.0893,
      "step": 12820
    },
    {
      "epoch": 0.9241692496215671,
      "grad_norm": 0.12386205391720449,
      "learning_rate": 3.0011308388769754e-06,
      "loss": 1.0137,
      "step": 12821
    },
    {
      "epoch": 0.924241332083904,
      "grad_norm": 0.12335550408843998,
      "learning_rate": 2.9954566220376514e-06,
      "loss": 1.0151,
      "step": 12822
    },
    {
      "epoch": 0.9243134145462409,
      "grad_norm": 0.18542272343386207,
      "learning_rate": 2.989787692813739e-06,
      "loss": 0.9801,
      "step": 12823
    },
    {
      "epoch": 0.9243854970085779,
      "grad_norm": 0.1341719167238453,
      "learning_rate": 2.984124051514292e-06,
      "loss": 1.0143,
      "step": 12824
    },
    {
      "epoch": 0.9244575794709148,
      "grad_norm": 0.10815034201712295,
      "learning_rate": 2.9784656984480074e-06,
      "loss": 1.0241,
      "step": 12825
    },
    {
      "epoch": 0.9245296619332516,
      "grad_norm": 0.1396925858001919,
      "learning_rate": 2.9728126339233385e-06,
      "loss": 1.0329,
      "step": 12826
    },
    {
      "epoch": 0.9246017443955885,
      "grad_norm": 0.13309004390235246,
      "learning_rate": 2.9671648582483945e-06,
      "loss": 1.0185,
      "step": 12827
    },
    {
      "epoch": 0.9246738268579254,
      "grad_norm": 0.12790292297158223,
      "learning_rate": 2.96152237173104e-06,
      "loss": 1.0794,
      "step": 12828
    },
    {
      "epoch": 0.9247459093202623,
      "grad_norm": 0.1194070353025491,
      "learning_rate": 2.9558851746788517e-06,
      "loss": 1.0152,
      "step": 12829
    },
    {
      "epoch": 0.9248179917825993,
      "grad_norm": 0.10951222307272522,
      "learning_rate": 2.9502532673991278e-06,
      "loss": 1.0813,
      "step": 12830
    },
    {
      "epoch": 0.9248900742449362,
      "grad_norm": 0.11377471833549095,
      "learning_rate": 2.944626650198812e-06,
      "loss": 0.9955,
      "step": 12831
    },
    {
      "epoch": 0.9249621567072731,
      "grad_norm": 0.12651266930441782,
      "learning_rate": 2.9390053233846136e-06,
      "loss": 1.0527,
      "step": 12832
    },
    {
      "epoch": 0.92503423916961,
      "grad_norm": 0.1221346072625863,
      "learning_rate": 2.9333892872629664e-06,
      "loss": 1.0868,
      "step": 12833
    },
    {
      "epoch": 0.925106321631947,
      "grad_norm": 0.12810855496713916,
      "learning_rate": 2.9277785421400027e-06,
      "loss": 1.0542,
      "step": 12834
    },
    {
      "epoch": 0.9251784040942839,
      "grad_norm": 0.1117708522984749,
      "learning_rate": 2.9221730883215337e-06,
      "loss": 0.9867,
      "step": 12835
    },
    {
      "epoch": 0.9252504865566208,
      "grad_norm": 0.12311325316671888,
      "learning_rate": 2.9165729261131035e-06,
      "loss": 1.0155,
      "step": 12836
    },
    {
      "epoch": 0.9253225690189577,
      "grad_norm": 0.1593908035098978,
      "learning_rate": 2.9109780558199794e-06,
      "loss": 1.0535,
      "step": 12837
    },
    {
      "epoch": 0.9253946514812946,
      "grad_norm": 0.12719914633490464,
      "learning_rate": 2.9053884777471506e-06,
      "loss": 0.9883,
      "step": 12838
    },
    {
      "epoch": 0.9254667339436315,
      "grad_norm": 0.16267680783127572,
      "learning_rate": 2.8998041921992735e-06,
      "loss": 0.9702,
      "step": 12839
    },
    {
      "epoch": 0.9255388164059685,
      "grad_norm": 0.13889169846526475,
      "learning_rate": 2.89422519948076e-06,
      "loss": 0.9576,
      "step": 12840
    },
    {
      "epoch": 0.9256108988683054,
      "grad_norm": 0.1270724620592686,
      "learning_rate": 2.8886514998956894e-06,
      "loss": 1.0268,
      "step": 12841
    },
    {
      "epoch": 0.9256829813306423,
      "grad_norm": 0.14894588717146956,
      "learning_rate": 2.883083093747907e-06,
      "loss": 1.0915,
      "step": 12842
    },
    {
      "epoch": 0.9257550637929791,
      "grad_norm": 0.11598312964575704,
      "learning_rate": 2.877519981340926e-06,
      "loss": 1.022,
      "step": 12843
    },
    {
      "epoch": 0.925827146255316,
      "grad_norm": 0.12245678544781287,
      "learning_rate": 2.8719621629779927e-06,
      "loss": 1.0198,
      "step": 12844
    },
    {
      "epoch": 0.925899228717653,
      "grad_norm": 0.1250064790729351,
      "learning_rate": 2.866409638962042e-06,
      "loss": 1.121,
      "step": 12845
    },
    {
      "epoch": 0.9259713111799899,
      "grad_norm": 0.1210061888367854,
      "learning_rate": 2.860862409595766e-06,
      "loss": 1.006,
      "step": 12846
    },
    {
      "epoch": 0.9260433936423268,
      "grad_norm": 0.11141707747608896,
      "learning_rate": 2.855320475181511e-06,
      "loss": 1.0558,
      "step": 12847
    },
    {
      "epoch": 0.9261154761046637,
      "grad_norm": 0.13021954082845869,
      "learning_rate": 2.849783836021369e-06,
      "loss": 1.0638,
      "step": 12848
    },
    {
      "epoch": 0.9261875585670006,
      "grad_norm": 0.12082824962192565,
      "learning_rate": 2.844252492417143e-06,
      "loss": 1.0553,
      "step": 12849
    },
    {
      "epoch": 0.9262596410293376,
      "grad_norm": 0.12626827441764665,
      "learning_rate": 2.8387264446703256e-06,
      "loss": 1.0425,
      "step": 12850
    },
    {
      "epoch": 0.9263317234916745,
      "grad_norm": 0.12314544897189109,
      "learning_rate": 2.833205693082164e-06,
      "loss": 1.0821,
      "step": 12851
    },
    {
      "epoch": 0.9264038059540114,
      "grad_norm": 0.1086297098261149,
      "learning_rate": 2.827690237953562e-06,
      "loss": 1.0158,
      "step": 12852
    },
    {
      "epoch": 0.9264758884163483,
      "grad_norm": 0.1087326785158667,
      "learning_rate": 2.822180079585168e-06,
      "loss": 1.0241,
      "step": 12853
    },
    {
      "epoch": 0.9265479708786852,
      "grad_norm": 0.14361397141051052,
      "learning_rate": 2.8166752182773315e-06,
      "loss": 1.004,
      "step": 12854
    },
    {
      "epoch": 0.9266200533410222,
      "grad_norm": 0.13823609048759125,
      "learning_rate": 2.811175654330134e-06,
      "loss": 1.0019,
      "step": 12855
    },
    {
      "epoch": 0.9266921358033591,
      "grad_norm": 0.13923630457673639,
      "learning_rate": 2.8056813880433354e-06,
      "loss": 1.0632,
      "step": 12856
    },
    {
      "epoch": 0.926764218265696,
      "grad_norm": 0.12324384611233052,
      "learning_rate": 2.80019241971643e-06,
      "loss": 1.0377,
      "step": 12857
    },
    {
      "epoch": 0.9268363007280329,
      "grad_norm": 0.15749268472331526,
      "learning_rate": 2.7947087496486e-06,
      "loss": 1.0162,
      "step": 12858
    },
    {
      "epoch": 0.9269083831903698,
      "grad_norm": 0.16361647398565185,
      "learning_rate": 2.7892303781387853e-06,
      "loss": 1.0351,
      "step": 12859
    },
    {
      "epoch": 0.9269804656527066,
      "grad_norm": 0.12928245608548145,
      "learning_rate": 2.783757305485568e-06,
      "loss": 1.1245,
      "step": 12860
    },
    {
      "epoch": 0.9270525481150436,
      "grad_norm": 0.11924869810182245,
      "learning_rate": 2.7782895319873105e-06,
      "loss": 1.0878,
      "step": 12861
    },
    {
      "epoch": 0.9271246305773805,
      "grad_norm": 0.1203377385905424,
      "learning_rate": 2.77282705794204e-06,
      "loss": 1.0796,
      "step": 12862
    },
    {
      "epoch": 0.9271967130397174,
      "grad_norm": 0.1218393956335747,
      "learning_rate": 2.767369883647519e-06,
      "loss": 1.044,
      "step": 12863
    },
    {
      "epoch": 0.9272687955020543,
      "grad_norm": 0.11310617887067208,
      "learning_rate": 2.7619180094011986e-06,
      "loss": 1.0452,
      "step": 12864
    },
    {
      "epoch": 0.9273408779643912,
      "grad_norm": 0.10971064045823717,
      "learning_rate": 2.756471435500274e-06,
      "loss": 1.0216,
      "step": 12865
    },
    {
      "epoch": 0.9274129604267282,
      "grad_norm": 0.11510535407089502,
      "learning_rate": 2.751030162241619e-06,
      "loss": 1.0183,
      "step": 12866
    },
    {
      "epoch": 0.9274850428890651,
      "grad_norm": 0.14178067040391493,
      "learning_rate": 2.7455941899218298e-06,
      "loss": 0.9636,
      "step": 12867
    },
    {
      "epoch": 0.927557125351402,
      "grad_norm": 0.12118923889881475,
      "learning_rate": 2.7401635188372244e-06,
      "loss": 0.976,
      "step": 12868
    },
    {
      "epoch": 0.9276292078137389,
      "grad_norm": 0.11374018932198057,
      "learning_rate": 2.7347381492838107e-06,
      "loss": 1.0006,
      "step": 12869
    },
    {
      "epoch": 0.9277012902760758,
      "grad_norm": 0.17896121580522384,
      "learning_rate": 2.7293180815573304e-06,
      "loss": 1.0357,
      "step": 12870
    },
    {
      "epoch": 0.9277733727384128,
      "grad_norm": 0.11951777018193624,
      "learning_rate": 2.7239033159532358e-06,
      "loss": 1.048,
      "step": 12871
    },
    {
      "epoch": 0.9278454552007497,
      "grad_norm": 0.18056699867556356,
      "learning_rate": 2.7184938527666457e-06,
      "loss": 1.0776,
      "step": 12872
    },
    {
      "epoch": 0.9279175376630866,
      "grad_norm": 0.11681875678861231,
      "learning_rate": 2.7130896922924587e-06,
      "loss": 1.0216,
      "step": 12873
    },
    {
      "epoch": 0.9279896201254235,
      "grad_norm": 0.12133755587270767,
      "learning_rate": 2.7076908348252383e-06,
      "loss": 1.0492,
      "step": 12874
    },
    {
      "epoch": 0.9280617025877604,
      "grad_norm": 0.12150497132945952,
      "learning_rate": 2.702297280659272e-06,
      "loss": 0.9973,
      "step": 12875
    },
    {
      "epoch": 0.9281337850500974,
      "grad_norm": 0.12474665266300103,
      "learning_rate": 2.6969090300885458e-06,
      "loss": 1.0784,
      "step": 12876
    },
    {
      "epoch": 0.9282058675124343,
      "grad_norm": 0.1525560590769466,
      "learning_rate": 2.69152608340677e-06,
      "loss": 0.9494,
      "step": 12877
    },
    {
      "epoch": 0.9282779499747711,
      "grad_norm": 0.1176616353028919,
      "learning_rate": 2.686148440907377e-06,
      "loss": 0.9995,
      "step": 12878
    },
    {
      "epoch": 0.928350032437108,
      "grad_norm": 0.1285419420423691,
      "learning_rate": 2.680776102883509e-06,
      "loss": 0.9819,
      "step": 12879
    },
    {
      "epoch": 0.9284221148994449,
      "grad_norm": 0.1244480766019781,
      "learning_rate": 2.6754090696279653e-06,
      "loss": 1.0505,
      "step": 12880
    },
    {
      "epoch": 0.9284941973617818,
      "grad_norm": 0.3214043342175789,
      "learning_rate": 2.6700473414333018e-06,
      "loss": 1.1034,
      "step": 12881
    },
    {
      "epoch": 0.9285662798241188,
      "grad_norm": 0.13166160225330611,
      "learning_rate": 2.664690918591828e-06,
      "loss": 0.9896,
      "step": 12882
    },
    {
      "epoch": 0.9286383622864557,
      "grad_norm": 0.12708296236665823,
      "learning_rate": 2.659339801395466e-06,
      "loss": 1.073,
      "step": 12883
    },
    {
      "epoch": 0.9287104447487926,
      "grad_norm": 0.13385304215285906,
      "learning_rate": 2.653993990135928e-06,
      "loss": 1.1565,
      "step": 12884
    },
    {
      "epoch": 0.9287825272111295,
      "grad_norm": 0.11898060999051976,
      "learning_rate": 2.6486534851045907e-06,
      "loss": 1.0113,
      "step": 12885
    },
    {
      "epoch": 0.9288546096734664,
      "grad_norm": 0.12747809430432222,
      "learning_rate": 2.6433182865925777e-06,
      "loss": 1.1127,
      "step": 12886
    },
    {
      "epoch": 0.9289266921358034,
      "grad_norm": 0.1434344786834499,
      "learning_rate": 2.6379883948907e-06,
      "loss": 1.0395,
      "step": 12887
    },
    {
      "epoch": 0.9289987745981403,
      "grad_norm": 0.11127786495193775,
      "learning_rate": 2.6326638102894706e-06,
      "loss": 1.0268,
      "step": 12888
    },
    {
      "epoch": 0.9290708570604772,
      "grad_norm": 0.13372843474237517,
      "learning_rate": 2.6273445330791234e-06,
      "loss": 0.985,
      "step": 12889
    },
    {
      "epoch": 0.9291429395228141,
      "grad_norm": 0.12013249379234839,
      "learning_rate": 2.622030563549649e-06,
      "loss": 1.0459,
      "step": 12890
    },
    {
      "epoch": 0.929215021985151,
      "grad_norm": 0.13791886341835385,
      "learning_rate": 2.6167219019906597e-06,
      "loss": 1.069,
      "step": 12891
    },
    {
      "epoch": 0.929287104447488,
      "grad_norm": 0.1264930347516963,
      "learning_rate": 2.6114185486915465e-06,
      "loss": 1.014,
      "step": 12892
    },
    {
      "epoch": 0.9293591869098249,
      "grad_norm": 0.11410247603354047,
      "learning_rate": 2.6061205039413893e-06,
      "loss": 1.0018,
      "step": 12893
    },
    {
      "epoch": 0.9294312693721618,
      "grad_norm": 0.11671576426988801,
      "learning_rate": 2.600827768028968e-06,
      "loss": 1.0363,
      "step": 12894
    },
    {
      "epoch": 0.9295033518344986,
      "grad_norm": 0.1254122100263261,
      "learning_rate": 2.595540341242797e-06,
      "loss": 0.9634,
      "step": 12895
    },
    {
      "epoch": 0.9295754342968355,
      "grad_norm": 0.12763983105430687,
      "learning_rate": 2.5902582238710783e-06,
      "loss": 1.0192,
      "step": 12896
    },
    {
      "epoch": 0.9296475167591725,
      "grad_norm": 0.11653640720145876,
      "learning_rate": 2.584981416201737e-06,
      "loss": 0.9922,
      "step": 12897
    },
    {
      "epoch": 0.9297195992215094,
      "grad_norm": 0.12624022672637344,
      "learning_rate": 2.5797099185224218e-06,
      "loss": 0.9732,
      "step": 12898
    },
    {
      "epoch": 0.9297916816838463,
      "grad_norm": 0.11757779557940816,
      "learning_rate": 2.574443731120446e-06,
      "loss": 1.1062,
      "step": 12899
    },
    {
      "epoch": 0.9298637641461832,
      "grad_norm": 0.1358122438824695,
      "learning_rate": 2.5691828542828922e-06,
      "loss": 1.1028,
      "step": 12900
    },
    {
      "epoch": 0.9299358466085201,
      "grad_norm": 0.11979328073044238,
      "learning_rate": 2.563927288296508e-06,
      "loss": 1.0094,
      "step": 12901
    },
    {
      "epoch": 0.930007929070857,
      "grad_norm": 0.11695712302350662,
      "learning_rate": 2.558677033447776e-06,
      "loss": 1.0431,
      "step": 12902
    },
    {
      "epoch": 0.930080011533194,
      "grad_norm": 0.12342029261860178,
      "learning_rate": 2.553432090022878e-06,
      "loss": 0.9523,
      "step": 12903
    },
    {
      "epoch": 0.9301520939955309,
      "grad_norm": 0.1098706487272026,
      "learning_rate": 2.548192458307719e-06,
      "loss": 0.9565,
      "step": 12904
    },
    {
      "epoch": 0.9302241764578678,
      "grad_norm": 0.1116620344363729,
      "learning_rate": 2.542958138587892e-06,
      "loss": 1.0287,
      "step": 12905
    },
    {
      "epoch": 0.9302962589202047,
      "grad_norm": 0.11254899103206681,
      "learning_rate": 2.537729131148725e-06,
      "loss": 0.9755,
      "step": 12906
    },
    {
      "epoch": 0.9303683413825417,
      "grad_norm": 0.11608271227141989,
      "learning_rate": 2.532505436275245e-06,
      "loss": 1.0948,
      "step": 12907
    },
    {
      "epoch": 0.9304404238448786,
      "grad_norm": 0.11671329276738265,
      "learning_rate": 2.527287054252192e-06,
      "loss": 1.002,
      "step": 12908
    },
    {
      "epoch": 0.9305125063072155,
      "grad_norm": 0.12094667506556418,
      "learning_rate": 2.522073985364004e-06,
      "loss": 0.9453,
      "step": 12909
    },
    {
      "epoch": 0.9305845887695524,
      "grad_norm": 0.14334938738917838,
      "learning_rate": 2.516866229894843e-06,
      "loss": 0.9965,
      "step": 12910
    },
    {
      "epoch": 0.9306566712318893,
      "grad_norm": 0.12990050179136575,
      "learning_rate": 2.511663788128582e-06,
      "loss": 1.0257,
      "step": 12911
    },
    {
      "epoch": 0.9307287536942261,
      "grad_norm": 0.11135851524503597,
      "learning_rate": 2.506466660348805e-06,
      "loss": 1.1114,
      "step": 12912
    },
    {
      "epoch": 0.9308008361565631,
      "grad_norm": 0.12075892096204867,
      "learning_rate": 2.501274846838797e-06,
      "loss": 0.952,
      "step": 12913
    },
    {
      "epoch": 0.9308729186189,
      "grad_norm": 0.11072685859975573,
      "learning_rate": 2.4960883478815642e-06,
      "loss": 1.001,
      "step": 12914
    },
    {
      "epoch": 0.9309450010812369,
      "grad_norm": 0.13365016716651693,
      "learning_rate": 2.490907163759815e-06,
      "loss": 1.1056,
      "step": 12915
    },
    {
      "epoch": 0.9310170835435738,
      "grad_norm": 0.15197249504952734,
      "learning_rate": 2.4857312947559443e-06,
      "loss": 1.0724,
      "step": 12916
    },
    {
      "epoch": 0.9310891660059107,
      "grad_norm": 0.12716930574704252,
      "learning_rate": 2.480560741152116e-06,
      "loss": 1.0668,
      "step": 12917
    },
    {
      "epoch": 0.9311612484682477,
      "grad_norm": 0.11425189160416473,
      "learning_rate": 2.47539550323016e-06,
      "loss": 1.0174,
      "step": 12918
    },
    {
      "epoch": 0.9312333309305846,
      "grad_norm": 0.15021310794568365,
      "learning_rate": 2.4702355812716404e-06,
      "loss": 1.0962,
      "step": 12919
    },
    {
      "epoch": 0.9313054133929215,
      "grad_norm": 0.12154149697056527,
      "learning_rate": 2.4650809755577873e-06,
      "loss": 1.0038,
      "step": 12920
    },
    {
      "epoch": 0.9313774958552584,
      "grad_norm": 0.11186456998453212,
      "learning_rate": 2.45993168636961e-06,
      "loss": 0.9884,
      "step": 12921
    },
    {
      "epoch": 0.9314495783175953,
      "grad_norm": 0.11868226983425761,
      "learning_rate": 2.4547877139877717e-06,
      "loss": 1.0742,
      "step": 12922
    },
    {
      "epoch": 0.9315216607799323,
      "grad_norm": 0.11713109906416624,
      "learning_rate": 2.4496490586926715e-06,
      "loss": 0.9823,
      "step": 12923
    },
    {
      "epoch": 0.9315937432422692,
      "grad_norm": 0.1278144254341914,
      "learning_rate": 2.4445157207644065e-06,
      "loss": 1.0566,
      "step": 12924
    },
    {
      "epoch": 0.9316658257046061,
      "grad_norm": 0.11484578797849854,
      "learning_rate": 2.439387700482776e-06,
      "loss": 0.9762,
      "step": 12925
    },
    {
      "epoch": 0.931737908166943,
      "grad_norm": 0.12016108768369564,
      "learning_rate": 2.4342649981273226e-06,
      "loss": 0.9983,
      "step": 12926
    },
    {
      "epoch": 0.9318099906292799,
      "grad_norm": 0.1257010398992546,
      "learning_rate": 2.4291476139772897e-06,
      "loss": 0.962,
      "step": 12927
    },
    {
      "epoch": 0.9318820730916169,
      "grad_norm": 0.1219348342099979,
      "learning_rate": 2.4240355483115982e-06,
      "loss": 1.0623,
      "step": 12928
    },
    {
      "epoch": 0.9319541555539538,
      "grad_norm": 0.1347026290715362,
      "learning_rate": 2.418928801408904e-06,
      "loss": 1.0455,
      "step": 12929
    },
    {
      "epoch": 0.9320262380162906,
      "grad_norm": 0.12929493334428802,
      "learning_rate": 2.4138273735475727e-06,
      "loss": 1.071,
      "step": 12930
    },
    {
      "epoch": 0.9320983204786275,
      "grad_norm": 0.1212100948762347,
      "learning_rate": 2.4087312650056926e-06,
      "loss": 1.0011,
      "step": 12931
    },
    {
      "epoch": 0.9321704029409644,
      "grad_norm": 0.12265969718179356,
      "learning_rate": 2.403640476061031e-06,
      "loss": 1.0379,
      "step": 12932
    },
    {
      "epoch": 0.9322424854033013,
      "grad_norm": 0.10873054669418636,
      "learning_rate": 2.3985550069910767e-06,
      "loss": 1.0327,
      "step": 12933
    },
    {
      "epoch": 0.9323145678656383,
      "grad_norm": 0.12250838996235335,
      "learning_rate": 2.3934748580730414e-06,
      "loss": 0.989,
      "step": 12934
    },
    {
      "epoch": 0.9323866503279752,
      "grad_norm": 0.13226764512991646,
      "learning_rate": 2.3884000295838593e-06,
      "loss": 0.9654,
      "step": 12935
    },
    {
      "epoch": 0.9324587327903121,
      "grad_norm": 0.12016387347032541,
      "learning_rate": 2.3833305218001202e-06,
      "loss": 1.036,
      "step": 12936
    },
    {
      "epoch": 0.932530815252649,
      "grad_norm": 0.11394881507791062,
      "learning_rate": 2.3782663349981693e-06,
      "loss": 1.0145,
      "step": 12937
    },
    {
      "epoch": 0.932602897714986,
      "grad_norm": 0.12925518637605618,
      "learning_rate": 2.3732074694540527e-06,
      "loss": 0.9965,
      "step": 12938
    },
    {
      "epoch": 0.9326749801773229,
      "grad_norm": 0.12358703271062894,
      "learning_rate": 2.368153925443528e-06,
      "loss": 1.0496,
      "step": 12939
    },
    {
      "epoch": 0.9327470626396598,
      "grad_norm": 0.11233152811502241,
      "learning_rate": 2.3631057032420523e-06,
      "loss": 1.0082,
      "step": 12940
    },
    {
      "epoch": 0.9328191451019967,
      "grad_norm": 0.11509306276917265,
      "learning_rate": 2.3580628031248054e-06,
      "loss": 1.106,
      "step": 12941
    },
    {
      "epoch": 0.9328912275643336,
      "grad_norm": 0.1300009986037204,
      "learning_rate": 2.353025225366645e-06,
      "loss": 1.0577,
      "step": 12942
    },
    {
      "epoch": 0.9329633100266705,
      "grad_norm": 0.12685060432809078,
      "learning_rate": 2.3479929702422077e-06,
      "loss": 1.0431,
      "step": 12943
    },
    {
      "epoch": 0.9330353924890075,
      "grad_norm": 0.14746374806064114,
      "learning_rate": 2.342966038025762e-06,
      "loss": 1.0499,
      "step": 12944
    },
    {
      "epoch": 0.9331074749513444,
      "grad_norm": 0.11765111913248838,
      "learning_rate": 2.3379444289913342e-06,
      "loss": 0.9541,
      "step": 12945
    },
    {
      "epoch": 0.9331795574136813,
      "grad_norm": 0.13577800170802976,
      "learning_rate": 2.3329281434126493e-06,
      "loss": 1.043,
      "step": 12946
    },
    {
      "epoch": 0.9332516398760181,
      "grad_norm": 0.1687818533991494,
      "learning_rate": 2.327917181563133e-06,
      "loss": 1.0467,
      "step": 12947
    },
    {
      "epoch": 0.933323722338355,
      "grad_norm": 0.11988785899527882,
      "learning_rate": 2.3229115437159223e-06,
      "loss": 1.0305,
      "step": 12948
    },
    {
      "epoch": 0.933395804800692,
      "grad_norm": 0.11686215432102774,
      "learning_rate": 2.3179112301438878e-06,
      "loss": 1.0595,
      "step": 12949
    },
    {
      "epoch": 0.9334678872630289,
      "grad_norm": 0.12069341316735605,
      "learning_rate": 2.312916241119567e-06,
      "loss": 1.0558,
      "step": 12950
    },
    {
      "epoch": 0.9335399697253658,
      "grad_norm": 0.10964273026969032,
      "learning_rate": 2.3079265769152536e-06,
      "loss": 1.0771,
      "step": 12951
    },
    {
      "epoch": 0.9336120521877027,
      "grad_norm": 0.18276490371261803,
      "learning_rate": 2.3029422378029184e-06,
      "loss": 0.9923,
      "step": 12952
    },
    {
      "epoch": 0.9336841346500396,
      "grad_norm": 0.12575735692261364,
      "learning_rate": 2.2979632240542557e-06,
      "loss": 1.0746,
      "step": 12953
    },
    {
      "epoch": 0.9337562171123766,
      "grad_norm": 0.1076299160622135,
      "learning_rate": 2.2929895359406595e-06,
      "loss": 1.0566,
      "step": 12954
    },
    {
      "epoch": 0.9338282995747135,
      "grad_norm": 0.1187184673372011,
      "learning_rate": 2.288021173733246e-06,
      "loss": 1.065,
      "step": 12955
    },
    {
      "epoch": 0.9339003820370504,
      "grad_norm": 0.12456030480978557,
      "learning_rate": 2.2830581377028424e-06,
      "loss": 1.004,
      "step": 12956
    },
    {
      "epoch": 0.9339724644993873,
      "grad_norm": 0.1229532039443307,
      "learning_rate": 2.278100428119967e-06,
      "loss": 1.0376,
      "step": 12957
    },
    {
      "epoch": 0.9340445469617242,
      "grad_norm": 0.1311700654544138,
      "learning_rate": 2.2731480452548582e-06,
      "loss": 0.9741,
      "step": 12958
    },
    {
      "epoch": 0.9341166294240612,
      "grad_norm": 0.10195492104982323,
      "learning_rate": 2.268200989377478e-06,
      "loss": 1.0358,
      "step": 12959
    },
    {
      "epoch": 0.9341887118863981,
      "grad_norm": 0.16683336934653917,
      "learning_rate": 2.2632592607574774e-06,
      "loss": 1.0785,
      "step": 12960
    },
    {
      "epoch": 0.934260794348735,
      "grad_norm": 0.1436015522723305,
      "learning_rate": 2.258322859664219e-06,
      "loss": 0.9096,
      "step": 12961
    },
    {
      "epoch": 0.9343328768110719,
      "grad_norm": 0.11236451755524539,
      "learning_rate": 2.2533917863667877e-06,
      "loss": 1.0216,
      "step": 12962
    },
    {
      "epoch": 0.9344049592734088,
      "grad_norm": 0.12978874403083382,
      "learning_rate": 2.2484660411339787e-06,
      "loss": 0.9424,
      "step": 12963
    },
    {
      "epoch": 0.9344770417357456,
      "grad_norm": 0.11819583789368085,
      "learning_rate": 2.2435456242342666e-06,
      "loss": 1.0704,
      "step": 12964
    },
    {
      "epoch": 0.9345491241980826,
      "grad_norm": 0.13245809239262168,
      "learning_rate": 2.2386305359358816e-06,
      "loss": 1.0786,
      "step": 12965
    },
    {
      "epoch": 0.9346212066604195,
      "grad_norm": 0.1320296070212614,
      "learning_rate": 2.2337207765067314e-06,
      "loss": 1.0527,
      "step": 12966
    },
    {
      "epoch": 0.9346932891227564,
      "grad_norm": 0.11549915981440365,
      "learning_rate": 2.2288163462144352e-06,
      "loss": 1.0603,
      "step": 12967
    },
    {
      "epoch": 0.9347653715850933,
      "grad_norm": 0.11650492049004701,
      "learning_rate": 2.223917245326357e-06,
      "loss": 1.0261,
      "step": 12968
    },
    {
      "epoch": 0.9348374540474302,
      "grad_norm": 0.12395188279776777,
      "learning_rate": 2.219023474109483e-06,
      "loss": 0.9975,
      "step": 12969
    },
    {
      "epoch": 0.9349095365097672,
      "grad_norm": 0.12027231799131939,
      "learning_rate": 2.2141350328306222e-06,
      "loss": 1.0758,
      "step": 12970
    },
    {
      "epoch": 0.9349816189721041,
      "grad_norm": 0.1904591507464848,
      "learning_rate": 2.2092519217562168e-06,
      "loss": 1.0133,
      "step": 12971
    },
    {
      "epoch": 0.935053701434441,
      "grad_norm": 0.11948862390446223,
      "learning_rate": 2.204374141152454e-06,
      "loss": 1.028,
      "step": 12972
    },
    {
      "epoch": 0.9351257838967779,
      "grad_norm": 0.12367851477429756,
      "learning_rate": 2.1995016912851775e-06,
      "loss": 1.0985,
      "step": 12973
    },
    {
      "epoch": 0.9351978663591148,
      "grad_norm": 0.11605756644345336,
      "learning_rate": 2.194634572420029e-06,
      "loss": 1.0425,
      "step": 12974
    },
    {
      "epoch": 0.9352699488214518,
      "grad_norm": 0.12080471137484668,
      "learning_rate": 2.189772784822286e-06,
      "loss": 1.0537,
      "step": 12975
    },
    {
      "epoch": 0.9353420312837887,
      "grad_norm": 0.12708539897633228,
      "learning_rate": 2.1849163287569586e-06,
      "loss": 1.0366,
      "step": 12976
    },
    {
      "epoch": 0.9354141137461256,
      "grad_norm": 0.11426335138583497,
      "learning_rate": 2.180065204488768e-06,
      "loss": 1.0188,
      "step": 12977
    },
    {
      "epoch": 0.9354861962084625,
      "grad_norm": 0.12181945680226752,
      "learning_rate": 2.175219412282148e-06,
      "loss": 0.9807,
      "step": 12978
    },
    {
      "epoch": 0.9355582786707994,
      "grad_norm": 0.12029843032939966,
      "learning_rate": 2.170378952401231e-06,
      "loss": 1.0489,
      "step": 12979
    },
    {
      "epoch": 0.9356303611331364,
      "grad_norm": 0.12970521202840007,
      "learning_rate": 2.165543825109884e-06,
      "loss": 1.0495,
      "step": 12980
    },
    {
      "epoch": 0.9357024435954733,
      "grad_norm": 0.21974202479877275,
      "learning_rate": 2.16071403067164e-06,
      "loss": 0.984,
      "step": 12981
    },
    {
      "epoch": 0.9357745260578101,
      "grad_norm": 0.12341585560577108,
      "learning_rate": 2.155889569349767e-06,
      "loss": 1.0237,
      "step": 12982
    },
    {
      "epoch": 0.935846608520147,
      "grad_norm": 0.12233904560326427,
      "learning_rate": 2.1510704414072656e-06,
      "loss": 1.114,
      "step": 12983
    },
    {
      "epoch": 0.9359186909824839,
      "grad_norm": 0.1224405498302888,
      "learning_rate": 2.1462566471068037e-06,
      "loss": 1.0816,
      "step": 12984
    },
    {
      "epoch": 0.9359907734448208,
      "grad_norm": 0.32649549384254933,
      "learning_rate": 2.141448186710782e-06,
      "loss": 1.1136,
      "step": 12985
    },
    {
      "epoch": 0.9360628559071578,
      "grad_norm": 0.15676570126646774,
      "learning_rate": 2.1366450604812905e-06,
      "loss": 1.0451,
      "step": 12986
    },
    {
      "epoch": 0.9361349383694947,
      "grad_norm": 0.11360107699731635,
      "learning_rate": 2.131847268680154e-06,
      "loss": 1.0346,
      "step": 12987
    },
    {
      "epoch": 0.9362070208318316,
      "grad_norm": 0.12374039844359348,
      "learning_rate": 2.127054811568907e-06,
      "loss": 1.0235,
      "step": 12988
    },
    {
      "epoch": 0.9362791032941685,
      "grad_norm": 0.112360610178571,
      "learning_rate": 2.1222676894087633e-06,
      "loss": 0.9797,
      "step": 12989
    },
    {
      "epoch": 0.9363511857565054,
      "grad_norm": 0.1498484772495191,
      "learning_rate": 2.11748590246067e-06,
      "loss": 1.0276,
      "step": 12990
    },
    {
      "epoch": 0.9364232682188424,
      "grad_norm": 0.12035891068832993,
      "learning_rate": 2.112709450985273e-06,
      "loss": 0.972,
      "step": 12991
    },
    {
      "epoch": 0.9364953506811793,
      "grad_norm": 0.11951505194968437,
      "learning_rate": 2.1079383352429536e-06,
      "loss": 0.9823,
      "step": 12992
    },
    {
      "epoch": 0.9365674331435162,
      "grad_norm": 0.12893334570691262,
      "learning_rate": 2.1031725554937377e-06,
      "loss": 1.0155,
      "step": 12993
    },
    {
      "epoch": 0.9366395156058531,
      "grad_norm": 0.10686347846360643,
      "learning_rate": 2.0984121119974386e-06,
      "loss": 1.0532,
      "step": 12994
    },
    {
      "epoch": 0.93671159806819,
      "grad_norm": 0.15500508627410733,
      "learning_rate": 2.0936570050135273e-06,
      "loss": 1.0162,
      "step": 12995
    },
    {
      "epoch": 0.936783680530527,
      "grad_norm": 0.1987931844697906,
      "learning_rate": 2.0889072348012184e-06,
      "loss": 1.0364,
      "step": 12996
    },
    {
      "epoch": 0.9368557629928639,
      "grad_norm": 0.1196546169016945,
      "learning_rate": 2.0841628016193936e-06,
      "loss": 0.9602,
      "step": 12997
    },
    {
      "epoch": 0.9369278454552008,
      "grad_norm": 0.12965929290859488,
      "learning_rate": 2.0794237057266796e-06,
      "loss": 1.0955,
      "step": 12998
    },
    {
      "epoch": 0.9369999279175376,
      "grad_norm": 0.41664004469527793,
      "learning_rate": 2.0746899473813807e-06,
      "loss": 1.0256,
      "step": 12999
    },
    {
      "epoch": 0.9370720103798745,
      "grad_norm": 0.12252865237546591,
      "learning_rate": 2.0699615268415684e-06,
      "loss": 1.0236,
      "step": 13000
    },
    {
      "epoch": 0.9371440928422115,
      "grad_norm": 0.12283385980736351,
      "learning_rate": 2.0652384443649474e-06,
      "loss": 1.1111,
      "step": 13001
    },
    {
      "epoch": 0.9372161753045484,
      "grad_norm": 0.12375842666652137,
      "learning_rate": 2.060520700208979e-06,
      "loss": 1.0379,
      "step": 13002
    },
    {
      "epoch": 0.9372882577668853,
      "grad_norm": 0.12064747233953219,
      "learning_rate": 2.0558082946308232e-06,
      "loss": 1.0147,
      "step": 13003
    },
    {
      "epoch": 0.9373603402292222,
      "grad_norm": 0.11234487477003002,
      "learning_rate": 2.051101227887353e-06,
      "loss": 0.9952,
      "step": 13004
    },
    {
      "epoch": 0.9374324226915591,
      "grad_norm": 0.11088260283999583,
      "learning_rate": 2.046399500235141e-06,
      "loss": 0.9976,
      "step": 13005
    },
    {
      "epoch": 0.937504505153896,
      "grad_norm": 0.12369908135664853,
      "learning_rate": 2.041703111930471e-06,
      "loss": 1.131,
      "step": 13006
    },
    {
      "epoch": 0.937576587616233,
      "grad_norm": 0.1293162807985883,
      "learning_rate": 2.037012063229349e-06,
      "loss": 0.9996,
      "step": 13007
    },
    {
      "epoch": 0.9376486700785699,
      "grad_norm": 0.1438924129186058,
      "learning_rate": 2.0323263543874592e-06,
      "loss": 1.0246,
      "step": 13008
    },
    {
      "epoch": 0.9377207525409068,
      "grad_norm": 0.12642398384816994,
      "learning_rate": 2.0276459856602424e-06,
      "loss": 1.0609,
      "step": 13009
    },
    {
      "epoch": 0.9377928350032437,
      "grad_norm": 0.12667881595061234,
      "learning_rate": 2.0229709573027943e-06,
      "loss": 1.0808,
      "step": 13010
    },
    {
      "epoch": 0.9378649174655806,
      "grad_norm": 0.12440621553965334,
      "learning_rate": 2.0183012695699554e-06,
      "loss": 0.9888,
      "step": 13011
    },
    {
      "epoch": 0.9379369999279176,
      "grad_norm": 0.11725585347278664,
      "learning_rate": 2.0136369227162777e-06,
      "loss": 1.0316,
      "step": 13012
    },
    {
      "epoch": 0.9380090823902545,
      "grad_norm": 0.1690349805720406,
      "learning_rate": 2.008977916995991e-06,
      "loss": 1.0223,
      "step": 13013
    },
    {
      "epoch": 0.9380811648525914,
      "grad_norm": 0.12517977082923612,
      "learning_rate": 2.004324252663059e-06,
      "loss": 1.0554,
      "step": 13014
    },
    {
      "epoch": 0.9381532473149283,
      "grad_norm": 0.1300642050158122,
      "learning_rate": 1.9996759299711567e-06,
      "loss": 1.0645,
      "step": 13015
    },
    {
      "epoch": 0.9382253297772651,
      "grad_norm": 0.12971931813358953,
      "learning_rate": 1.9950329491736476e-06,
      "loss": 0.9862,
      "step": 13016
    },
    {
      "epoch": 0.9382974122396021,
      "grad_norm": 0.11817754786615772,
      "learning_rate": 1.9903953105236186e-06,
      "loss": 1.0248,
      "step": 13017
    },
    {
      "epoch": 0.938369494701939,
      "grad_norm": 0.13698280900255264,
      "learning_rate": 1.985763014273867e-06,
      "loss": 1.0235,
      "step": 13018
    },
    {
      "epoch": 0.9384415771642759,
      "grad_norm": 0.13974044475934372,
      "learning_rate": 1.9811360606768805e-06,
      "loss": 1.0602,
      "step": 13019
    },
    {
      "epoch": 0.9385136596266128,
      "grad_norm": 0.10861473232741037,
      "learning_rate": 1.97651444998489e-06,
      "loss": 1.016,
      "step": 13020
    },
    {
      "epoch": 0.9385857420889497,
      "grad_norm": 0.12243977419019622,
      "learning_rate": 1.971898182449805e-06,
      "loss": 0.9584,
      "step": 13021
    },
    {
      "epoch": 0.9386578245512867,
      "grad_norm": 0.12788786198636445,
      "learning_rate": 1.967287258323258e-06,
      "loss": 1.0255,
      "step": 13022
    },
    {
      "epoch": 0.9387299070136236,
      "grad_norm": 0.127413241879665,
      "learning_rate": 1.9626816778565813e-06,
      "loss": 1.1034,
      "step": 13023
    },
    {
      "epoch": 0.9388019894759605,
      "grad_norm": 0.10732903779325993,
      "learning_rate": 1.9580814413008075e-06,
      "loss": 1.1128,
      "step": 13024
    },
    {
      "epoch": 0.9388740719382974,
      "grad_norm": 0.12823605008000982,
      "learning_rate": 1.9534865489067246e-06,
      "loss": 1.0576,
      "step": 13025
    },
    {
      "epoch": 0.9389461544006343,
      "grad_norm": 0.11069758446445412,
      "learning_rate": 1.9488970009247544e-06,
      "loss": 1.0983,
      "step": 13026
    },
    {
      "epoch": 0.9390182368629713,
      "grad_norm": 0.12080617719942739,
      "learning_rate": 1.944312797605097e-06,
      "loss": 1.0923,
      "step": 13027
    },
    {
      "epoch": 0.9390903193253082,
      "grad_norm": 0.12553623694997823,
      "learning_rate": 1.939733939197619e-06,
      "loss": 1.0778,
      "step": 13028
    },
    {
      "epoch": 0.9391624017876451,
      "grad_norm": 0.1437969394987447,
      "learning_rate": 1.9351604259519206e-06,
      "loss": 1.0087,
      "step": 13029
    },
    {
      "epoch": 0.939234484249982,
      "grad_norm": 0.17362685559864038,
      "learning_rate": 1.930592258117281e-06,
      "loss": 1.0543,
      "step": 13030
    },
    {
      "epoch": 0.9393065667123189,
      "grad_norm": 0.12823531224861653,
      "learning_rate": 1.9260294359427223e-06,
      "loss": 1.0207,
      "step": 13031
    },
    {
      "epoch": 0.9393786491746559,
      "grad_norm": 0.11924286685984527,
      "learning_rate": 1.921471959676957e-06,
      "loss": 1.0203,
      "step": 13032
    },
    {
      "epoch": 0.9394507316369927,
      "grad_norm": 0.6921005358379819,
      "learning_rate": 1.9169198295684087e-06,
      "loss": 1.0736,
      "step": 13033
    },
    {
      "epoch": 0.9395228140993296,
      "grad_norm": 0.10734054079139241,
      "learning_rate": 1.912373045865201e-06,
      "loss": 1.0534,
      "step": 13034
    },
    {
      "epoch": 0.9395948965616665,
      "grad_norm": 0.12384481582828599,
      "learning_rate": 1.9078316088151693e-06,
      "loss": 1.1017,
      "step": 13035
    },
    {
      "epoch": 0.9396669790240034,
      "grad_norm": 0.1140835660572683,
      "learning_rate": 1.903295518665893e-06,
      "loss": 0.9804,
      "step": 13036
    },
    {
      "epoch": 0.9397390614863403,
      "grad_norm": 0.12325830916229079,
      "learning_rate": 1.8987647756645966e-06,
      "loss": 1.0106,
      "step": 13037
    },
    {
      "epoch": 0.9398111439486773,
      "grad_norm": 2.0063796761535437,
      "learning_rate": 1.8942393800582603e-06,
      "loss": 0.989,
      "step": 13038
    },
    {
      "epoch": 0.9398832264110142,
      "grad_norm": 0.12871694991930682,
      "learning_rate": 1.8897193320935536e-06,
      "loss": 0.9742,
      "step": 13039
    },
    {
      "epoch": 0.9399553088733511,
      "grad_norm": 0.12805789952994576,
      "learning_rate": 1.885204632016868e-06,
      "loss": 1.0512,
      "step": 13040
    },
    {
      "epoch": 0.940027391335688,
      "grad_norm": 0.12030847973106878,
      "learning_rate": 1.8806952800742849e-06,
      "loss": 1.0298,
      "step": 13041
    },
    {
      "epoch": 0.9400994737980249,
      "grad_norm": 0.1568588979248864,
      "learning_rate": 1.8761912765116185e-06,
      "loss": 0.9826,
      "step": 13042
    },
    {
      "epoch": 0.9401715562603619,
      "grad_norm": 0.11940736205950533,
      "learning_rate": 1.8716926215743612e-06,
      "loss": 0.9968,
      "step": 13043
    },
    {
      "epoch": 0.9402436387226988,
      "grad_norm": 0.1180767448585996,
      "learning_rate": 1.8671993155077393e-06,
      "loss": 1.04,
      "step": 13044
    },
    {
      "epoch": 0.9403157211850357,
      "grad_norm": 0.12575085604848768,
      "learning_rate": 1.862711358556679e-06,
      "loss": 1.0345,
      "step": 13045
    },
    {
      "epoch": 0.9403878036473726,
      "grad_norm": 0.11706913306163731,
      "learning_rate": 1.858228750965807e-06,
      "loss": 1.0599,
      "step": 13046
    },
    {
      "epoch": 0.9404598861097095,
      "grad_norm": 0.1288215415894023,
      "learning_rate": 1.8537514929794608e-06,
      "loss": 1.0959,
      "step": 13047
    },
    {
      "epoch": 0.9405319685720465,
      "grad_norm": 0.14136405532074578,
      "learning_rate": 1.8492795848417122e-06,
      "loss": 1.0042,
      "step": 13048
    },
    {
      "epoch": 0.9406040510343834,
      "grad_norm": 0.14021521128952058,
      "learning_rate": 1.8448130267962992e-06,
      "loss": 1.0316,
      "step": 13049
    },
    {
      "epoch": 0.9406761334967203,
      "grad_norm": 0.13210879590101088,
      "learning_rate": 1.8403518190867053e-06,
      "loss": 0.9975,
      "step": 13050
    },
    {
      "epoch": 0.9407482159590571,
      "grad_norm": 0.14509484044000456,
      "learning_rate": 1.8358959619560911e-06,
      "loss": 0.9648,
      "step": 13051
    },
    {
      "epoch": 0.940820298421394,
      "grad_norm": 0.11634408515332993,
      "learning_rate": 1.8314454556473625e-06,
      "loss": 1.0358,
      "step": 13052
    },
    {
      "epoch": 0.940892380883731,
      "grad_norm": 0.2093863770856058,
      "learning_rate": 1.827000300403081e-06,
      "loss": 1.0659,
      "step": 13053
    },
    {
      "epoch": 0.9409644633460679,
      "grad_norm": 0.11323764704080268,
      "learning_rate": 1.8225604964655752e-06,
      "loss": 1.051,
      "step": 13054
    },
    {
      "epoch": 0.9410365458084048,
      "grad_norm": 0.1171093368803529,
      "learning_rate": 1.81812604407684e-06,
      "loss": 1.0623,
      "step": 13055
    },
    {
      "epoch": 0.9411086282707417,
      "grad_norm": 0.13898819966102985,
      "learning_rate": 1.8136969434785932e-06,
      "loss": 1.0648,
      "step": 13056
    },
    {
      "epoch": 0.9411807107330786,
      "grad_norm": 0.11878719222243925,
      "learning_rate": 1.8092731949122532e-06,
      "loss": 1.0098,
      "step": 13057
    },
    {
      "epoch": 0.9412527931954155,
      "grad_norm": 0.747711923045986,
      "learning_rate": 1.8048547986189822e-06,
      "loss": 1.0362,
      "step": 13058
    },
    {
      "epoch": 0.9413248756577525,
      "grad_norm": 0.11677887978101839,
      "learning_rate": 1.800441754839588e-06,
      "loss": 1.1018,
      "step": 13059
    },
    {
      "epoch": 0.9413969581200894,
      "grad_norm": 0.13152449319206932,
      "learning_rate": 1.7960340638146444e-06,
      "loss": 0.9895,
      "step": 13060
    },
    {
      "epoch": 0.9414690405824263,
      "grad_norm": 0.10477712413731434,
      "learning_rate": 1.7916317257844039e-06,
      "loss": 1.0187,
      "step": 13061
    },
    {
      "epoch": 0.9415411230447632,
      "grad_norm": 0.12364586923741917,
      "learning_rate": 1.7872347409888301e-06,
      "loss": 1.128,
      "step": 13062
    },
    {
      "epoch": 0.9416132055071001,
      "grad_norm": 0.18108628885160144,
      "learning_rate": 1.782843109667598e-06,
      "loss": 1.0068,
      "step": 13063
    },
    {
      "epoch": 0.9416852879694371,
      "grad_norm": 0.12999161236172285,
      "learning_rate": 1.7784568320600825e-06,
      "loss": 1.078,
      "step": 13064
    },
    {
      "epoch": 0.941757370431774,
      "grad_norm": 0.1749022265859241,
      "learning_rate": 1.7740759084054037e-06,
      "loss": 0.9985,
      "step": 13065
    },
    {
      "epoch": 0.9418294528941109,
      "grad_norm": 0.14897912444463438,
      "learning_rate": 1.7697003389423262e-06,
      "loss": 1.0349,
      "step": 13066
    },
    {
      "epoch": 0.9419015353564478,
      "grad_norm": 0.13476966888714143,
      "learning_rate": 1.765330123909381e-06,
      "loss": 1.0218,
      "step": 13067
    },
    {
      "epoch": 0.9419736178187846,
      "grad_norm": 0.12116631917302322,
      "learning_rate": 1.7609652635447782e-06,
      "loss": 0.9817,
      "step": 13068
    },
    {
      "epoch": 0.9420457002811216,
      "grad_norm": 0.11284771256940551,
      "learning_rate": 1.7566057580864382e-06,
      "loss": 0.9758,
      "step": 13069
    },
    {
      "epoch": 0.9421177827434585,
      "grad_norm": 0.11538615788417012,
      "learning_rate": 1.752251607771993e-06,
      "loss": 1.0096,
      "step": 13070
    },
    {
      "epoch": 0.9421898652057954,
      "grad_norm": 0.11552315162515048,
      "learning_rate": 1.7479028128387865e-06,
      "loss": 1.0844,
      "step": 13071
    },
    {
      "epoch": 0.9422619476681323,
      "grad_norm": 0.13662522235289076,
      "learning_rate": 1.743559373523862e-06,
      "loss": 1.0259,
      "step": 13072
    },
    {
      "epoch": 0.9423340301304692,
      "grad_norm": 0.12462782917734105,
      "learning_rate": 1.7392212900639859e-06,
      "loss": 0.9828,
      "step": 13073
    },
    {
      "epoch": 0.9424061125928062,
      "grad_norm": 0.1221245365478055,
      "learning_rate": 1.7348885626956024e-06,
      "loss": 1.0385,
      "step": 13074
    },
    {
      "epoch": 0.9424781950551431,
      "grad_norm": 0.1387634436174156,
      "learning_rate": 1.7305611916549114e-06,
      "loss": 1.0056,
      "step": 13075
    },
    {
      "epoch": 0.94255027751748,
      "grad_norm": 0.11112786233421847,
      "learning_rate": 1.7262391771777687e-06,
      "loss": 0.9715,
      "step": 13076
    },
    {
      "epoch": 0.9426223599798169,
      "grad_norm": 0.1298143999653855,
      "learning_rate": 1.7219225194997857e-06,
      "loss": 1.097,
      "step": 13077
    },
    {
      "epoch": 0.9426944424421538,
      "grad_norm": 0.11303764771351621,
      "learning_rate": 1.717611218856241e-06,
      "loss": 1.0568,
      "step": 13078
    },
    {
      "epoch": 0.9427665249044908,
      "grad_norm": 0.31434739392806416,
      "learning_rate": 1.7133052754821355e-06,
      "loss": 1.065,
      "step": 13079
    },
    {
      "epoch": 0.9428386073668277,
      "grad_norm": 0.13003257010317396,
      "learning_rate": 1.7090046896121925e-06,
      "loss": 1.0565,
      "step": 13080
    },
    {
      "epoch": 0.9429106898291646,
      "grad_norm": 0.11485867513678374,
      "learning_rate": 1.7047094614808357e-06,
      "loss": 1.0485,
      "step": 13081
    },
    {
      "epoch": 0.9429827722915015,
      "grad_norm": 0.12218177001198599,
      "learning_rate": 1.7004195913221888e-06,
      "loss": 1.1145,
      "step": 13082
    },
    {
      "epoch": 0.9430548547538384,
      "grad_norm": 1.32698855784716,
      "learning_rate": 1.696135079370076e-06,
      "loss": 0.9929,
      "step": 13083
    },
    {
      "epoch": 0.9431269372161754,
      "grad_norm": 0.1305756486412518,
      "learning_rate": 1.6918559258580657e-06,
      "loss": 1.0771,
      "step": 13084
    },
    {
      "epoch": 0.9431990196785122,
      "grad_norm": 0.11711169684566826,
      "learning_rate": 1.687582131019394e-06,
      "loss": 1.0643,
      "step": 13085
    },
    {
      "epoch": 0.9432711021408491,
      "grad_norm": 0.13149536132106446,
      "learning_rate": 1.68331369508703e-06,
      "loss": 1.1399,
      "step": 13086
    },
    {
      "epoch": 0.943343184603186,
      "grad_norm": 0.12476295000138912,
      "learning_rate": 1.6790506182936316e-06,
      "loss": 0.9725,
      "step": 13087
    },
    {
      "epoch": 0.9434152670655229,
      "grad_norm": 0.3515139284731719,
      "learning_rate": 1.674792900871569e-06,
      "loss": 1.0162,
      "step": 13088
    },
    {
      "epoch": 0.9434873495278598,
      "grad_norm": 0.2523050756542362,
      "learning_rate": 1.6705405430529452e-06,
      "loss": 1.0359,
      "step": 13089
    },
    {
      "epoch": 0.9435594319901968,
      "grad_norm": 0.1148559032232396,
      "learning_rate": 1.6662935450695416e-06,
      "loss": 1.0189,
      "step": 13090
    },
    {
      "epoch": 0.9436315144525337,
      "grad_norm": 0.14304473509678312,
      "learning_rate": 1.6620519071528506e-06,
      "loss": 1.0853,
      "step": 13091
    },
    {
      "epoch": 0.9437035969148706,
      "grad_norm": 0.14111112074786084,
      "learning_rate": 1.6578156295340874e-06,
      "loss": 1.1183,
      "step": 13092
    },
    {
      "epoch": 0.9437756793772075,
      "grad_norm": 0.14143197591220513,
      "learning_rate": 1.6535847124441783e-06,
      "loss": 1.037,
      "step": 13093
    },
    {
      "epoch": 0.9438477618395444,
      "grad_norm": 0.12773753979816568,
      "learning_rate": 1.6493591561137388e-06,
      "loss": 1.0544,
      "step": 13094
    },
    {
      "epoch": 0.9439198443018814,
      "grad_norm": 0.112575108987566,
      "learning_rate": 1.6451389607730849e-06,
      "loss": 1.0271,
      "step": 13095
    },
    {
      "epoch": 0.9439919267642183,
      "grad_norm": 0.1342701449104721,
      "learning_rate": 1.6409241266522546e-06,
      "loss": 1.0934,
      "step": 13096
    },
    {
      "epoch": 0.9440640092265552,
      "grad_norm": 0.12893623972251575,
      "learning_rate": 1.6367146539810196e-06,
      "loss": 0.9928,
      "step": 13097
    },
    {
      "epoch": 0.9441360916888921,
      "grad_norm": 0.12680566588464073,
      "learning_rate": 1.6325105429888188e-06,
      "loss": 1.0633,
      "step": 13098
    },
    {
      "epoch": 0.944208174151229,
      "grad_norm": 0.12891487570709392,
      "learning_rate": 1.6283117939048131e-06,
      "loss": 1.0535,
      "step": 13099
    },
    {
      "epoch": 0.944280256613566,
      "grad_norm": 0.12590015879024843,
      "learning_rate": 1.6241184069578641e-06,
      "loss": 1.0687,
      "step": 13100
    },
    {
      "epoch": 0.9443523390759029,
      "grad_norm": 0.12326792417461893,
      "learning_rate": 1.6199303823765665e-06,
      "loss": 1.0022,
      "step": 13101
    },
    {
      "epoch": 0.9444244215382398,
      "grad_norm": 0.4831468145178327,
      "learning_rate": 1.6157477203891936e-06,
      "loss": 1.0589,
      "step": 13102
    },
    {
      "epoch": 0.9444965040005766,
      "grad_norm": 0.1149181503114315,
      "learning_rate": 1.6115704212237403e-06,
      "loss": 1.0037,
      "step": 13103
    },
    {
      "epoch": 0.9445685864629135,
      "grad_norm": 0.11510199231515798,
      "learning_rate": 1.6073984851079139e-06,
      "loss": 1.0381,
      "step": 13104
    },
    {
      "epoch": 0.9446406689252504,
      "grad_norm": 0.13250789993301515,
      "learning_rate": 1.6032319122691097e-06,
      "loss": 1.0397,
      "step": 13105
    },
    {
      "epoch": 0.9447127513875874,
      "grad_norm": 0.1562424112159372,
      "learning_rate": 1.5990707029344464e-06,
      "loss": 1.0303,
      "step": 13106
    },
    {
      "epoch": 0.9447848338499243,
      "grad_norm": 0.14489789862499122,
      "learning_rate": 1.5949148573307538e-06,
      "loss": 1.0695,
      "step": 13107
    },
    {
      "epoch": 0.9448569163122612,
      "grad_norm": 0.11048731526393082,
      "learning_rate": 1.5907643756845503e-06,
      "loss": 0.9496,
      "step": 13108
    },
    {
      "epoch": 0.9449289987745981,
      "grad_norm": 0.1386756048671415,
      "learning_rate": 1.5866192582220884e-06,
      "loss": 1.0476,
      "step": 13109
    },
    {
      "epoch": 0.945001081236935,
      "grad_norm": 0.1370886851790602,
      "learning_rate": 1.5824795051693098e-06,
      "loss": 1.0501,
      "step": 13110
    },
    {
      "epoch": 0.945073163699272,
      "grad_norm": 0.10579745552821532,
      "learning_rate": 1.578345116751867e-06,
      "loss": 1.1801,
      "step": 13111
    },
    {
      "epoch": 0.9451452461616089,
      "grad_norm": 0.13877406930688901,
      "learning_rate": 1.5742160931951133e-06,
      "loss": 1.019,
      "step": 13112
    },
    {
      "epoch": 0.9452173286239458,
      "grad_norm": 0.1260145549369598,
      "learning_rate": 1.5700924347241353e-06,
      "loss": 1.0101,
      "step": 13113
    },
    {
      "epoch": 0.9452894110862827,
      "grad_norm": 0.22016158947347228,
      "learning_rate": 1.5659741415636975e-06,
      "loss": 1.0873,
      "step": 13114
    },
    {
      "epoch": 0.9453614935486196,
      "grad_norm": 0.12240629682899319,
      "learning_rate": 1.561861213938276e-06,
      "loss": 0.991,
      "step": 13115
    },
    {
      "epoch": 0.9454335760109566,
      "grad_norm": 0.14069067472993388,
      "learning_rate": 1.5577536520720803e-06,
      "loss": 1.1107,
      "step": 13116
    },
    {
      "epoch": 0.9455056584732935,
      "grad_norm": 0.1188179690573619,
      "learning_rate": 1.5536514561889982e-06,
      "loss": 1.0695,
      "step": 13117
    },
    {
      "epoch": 0.9455777409356304,
      "grad_norm": 0.13762180188928616,
      "learning_rate": 1.5495546265126283e-06,
      "loss": 1.0266,
      "step": 13118
    },
    {
      "epoch": 0.9456498233979673,
      "grad_norm": 0.10916034076545979,
      "learning_rate": 1.545463163266303e-06,
      "loss": 1.0397,
      "step": 13119
    },
    {
      "epoch": 0.9457219058603041,
      "grad_norm": 0.12192366772425679,
      "learning_rate": 1.5413770666730331e-06,
      "loss": 0.976,
      "step": 13120
    },
    {
      "epoch": 0.945793988322641,
      "grad_norm": 0.12754656514705517,
      "learning_rate": 1.5372963369555516e-06,
      "loss": 1.0238,
      "step": 13121
    },
    {
      "epoch": 0.945866070784978,
      "grad_norm": 0.1184023699554701,
      "learning_rate": 1.5332209743362914e-06,
      "loss": 1.0263,
      "step": 13122
    },
    {
      "epoch": 0.9459381532473149,
      "grad_norm": 0.12895133903345699,
      "learning_rate": 1.529150979037397e-06,
      "loss": 1.0625,
      "step": 13123
    },
    {
      "epoch": 0.9460102357096518,
      "grad_norm": 0.13347910081147496,
      "learning_rate": 1.5250863512807134e-06,
      "loss": 1.1018,
      "step": 13124
    },
    {
      "epoch": 0.9460823181719887,
      "grad_norm": 0.11289121099812395,
      "learning_rate": 1.5210270912878077e-06,
      "loss": 0.9658,
      "step": 13125
    },
    {
      "epoch": 0.9461544006343257,
      "grad_norm": 0.11863895225751596,
      "learning_rate": 1.5169731992799585e-06,
      "loss": 1.12,
      "step": 13126
    },
    {
      "epoch": 0.9462264830966626,
      "grad_norm": 0.13388342084374266,
      "learning_rate": 1.5129246754781001e-06,
      "loss": 0.9859,
      "step": 13127
    },
    {
      "epoch": 0.9462985655589995,
      "grad_norm": 0.1413247807191494,
      "learning_rate": 1.508881520102956e-06,
      "loss": 0.9969,
      "step": 13128
    },
    {
      "epoch": 0.9463706480213364,
      "grad_norm": 0.12856107071960687,
      "learning_rate": 1.5048437333748833e-06,
      "loss": 1.0709,
      "step": 13129
    },
    {
      "epoch": 0.9464427304836733,
      "grad_norm": 0.12439310496498707,
      "learning_rate": 1.5008113155140058e-06,
      "loss": 1.0586,
      "step": 13130
    },
    {
      "epoch": 0.9465148129460103,
      "grad_norm": 0.1248956650699333,
      "learning_rate": 1.4967842667401034e-06,
      "loss": 1.0595,
      "step": 13131
    },
    {
      "epoch": 0.9465868954083472,
      "grad_norm": 1.0556052161588956,
      "learning_rate": 1.492762587272678e-06,
      "loss": 1.01,
      "step": 13132
    },
    {
      "epoch": 0.9466589778706841,
      "grad_norm": 0.14108119635035504,
      "learning_rate": 1.4887462773309767e-06,
      "loss": 1.049,
      "step": 13133
    },
    {
      "epoch": 0.946731060333021,
      "grad_norm": 0.1444444777500677,
      "learning_rate": 1.4847353371339135e-06,
      "loss": 1.0324,
      "step": 13134
    },
    {
      "epoch": 0.9468031427953579,
      "grad_norm": 0.1138891192899385,
      "learning_rate": 1.480729766900113e-06,
      "loss": 1.0224,
      "step": 13135
    },
    {
      "epoch": 0.9468752252576949,
      "grad_norm": 0.1222874704872856,
      "learning_rate": 1.476729566847912e-06,
      "loss": 1.0671,
      "step": 13136
    },
    {
      "epoch": 0.9469473077200317,
      "grad_norm": 0.11483772920363128,
      "learning_rate": 1.4727347371953692e-06,
      "loss": 0.9655,
      "step": 13137
    },
    {
      "epoch": 0.9470193901823686,
      "grad_norm": 0.1162722980308502,
      "learning_rate": 1.468745278160244e-06,
      "loss": 1.0292,
      "step": 13138
    },
    {
      "epoch": 0.9470914726447055,
      "grad_norm": 0.12002167753993222,
      "learning_rate": 1.4647611899599734e-06,
      "loss": 1.0515,
      "step": 13139
    },
    {
      "epoch": 0.9471635551070424,
      "grad_norm": 0.13290058211879016,
      "learning_rate": 1.4607824728117392e-06,
      "loss": 1.1051,
      "step": 13140
    },
    {
      "epoch": 0.9472356375693793,
      "grad_norm": 0.13343341372396697,
      "learning_rate": 1.4568091269324124e-06,
      "loss": 1.0677,
      "step": 13141
    },
    {
      "epoch": 0.9473077200317163,
      "grad_norm": 0.12470501482726587,
      "learning_rate": 1.4528411525385976e-06,
      "loss": 1.0706,
      "step": 13142
    },
    {
      "epoch": 0.9473798024940532,
      "grad_norm": 0.146772617563711,
      "learning_rate": 1.4488785498465663e-06,
      "loss": 1.0268,
      "step": 13143
    },
    {
      "epoch": 0.9474518849563901,
      "grad_norm": 0.5093429559688946,
      "learning_rate": 1.4449213190723121e-06,
      "loss": 1.0912,
      "step": 13144
    },
    {
      "epoch": 0.947523967418727,
      "grad_norm": 0.24380523571029256,
      "learning_rate": 1.4409694604315294e-06,
      "loss": 1.0406,
      "step": 13145
    },
    {
      "epoch": 0.9475960498810639,
      "grad_norm": 0.12945254832960723,
      "learning_rate": 1.437022974139668e-06,
      "loss": 1.0124,
      "step": 13146
    },
    {
      "epoch": 0.9476681323434009,
      "grad_norm": 0.13648759887021059,
      "learning_rate": 1.4330818604118113e-06,
      "loss": 1.0439,
      "step": 13147
    },
    {
      "epoch": 0.9477402148057378,
      "grad_norm": 0.1333669230862397,
      "learning_rate": 1.4291461194628098e-06,
      "loss": 1.0114,
      "step": 13148
    },
    {
      "epoch": 0.9478122972680747,
      "grad_norm": 0.16149078303767267,
      "learning_rate": 1.4252157515071697e-06,
      "loss": 1.0163,
      "step": 13149
    },
    {
      "epoch": 0.9478843797304116,
      "grad_norm": 0.13007502441998878,
      "learning_rate": 1.4212907567591638e-06,
      "loss": 1.0014,
      "step": 13150
    },
    {
      "epoch": 0.9479564621927485,
      "grad_norm": 0.11282209541523977,
      "learning_rate": 1.4173711354327101e-06,
      "loss": 0.9516,
      "step": 13151
    },
    {
      "epoch": 0.9480285446550855,
      "grad_norm": 0.12193358053130046,
      "learning_rate": 1.4134568877414711e-06,
      "loss": 1.0637,
      "step": 13152
    },
    {
      "epoch": 0.9481006271174224,
      "grad_norm": 0.13290503489229663,
      "learning_rate": 1.4095480138988203e-06,
      "loss": 0.992,
      "step": 13153
    },
    {
      "epoch": 0.9481727095797593,
      "grad_norm": 0.14681371896374487,
      "learning_rate": 1.4056445141178208e-06,
      "loss": 1.0666,
      "step": 13154
    },
    {
      "epoch": 0.9482447920420961,
      "grad_norm": 0.11933992887922519,
      "learning_rate": 1.401746388611236e-06,
      "loss": 1.1158,
      "step": 13155
    },
    {
      "epoch": 0.948316874504433,
      "grad_norm": 0.11760852335943557,
      "learning_rate": 1.397853637591562e-06,
      "loss": 1.0419,
      "step": 13156
    },
    {
      "epoch": 0.94838895696677,
      "grad_norm": 0.10607668384273662,
      "learning_rate": 1.393966261270996e-06,
      "loss": 1.0451,
      "step": 13157
    },
    {
      "epoch": 0.9484610394291069,
      "grad_norm": 0.12597935698831939,
      "learning_rate": 1.3900842598614128e-06,
      "loss": 0.9781,
      "step": 13158
    },
    {
      "epoch": 0.9485331218914438,
      "grad_norm": 0.11802286199841613,
      "learning_rate": 1.3862076335744211e-06,
      "loss": 1.0785,
      "step": 13159
    },
    {
      "epoch": 0.9486052043537807,
      "grad_norm": 0.1640912584327454,
      "learning_rate": 1.382336382621341e-06,
      "loss": 0.9971,
      "step": 13160
    },
    {
      "epoch": 0.9486772868161176,
      "grad_norm": 0.12462851129613037,
      "learning_rate": 1.3784705072131921e-06,
      "loss": 1.047,
      "step": 13161
    },
    {
      "epoch": 0.9487493692784545,
      "grad_norm": 0.14778887739978674,
      "learning_rate": 1.3746100075606949e-06,
      "loss": 1.1127,
      "step": 13162
    },
    {
      "epoch": 0.9488214517407915,
      "grad_norm": 0.15112146785931027,
      "learning_rate": 1.3707548838742811e-06,
      "loss": 0.9988,
      "step": 13163
    },
    {
      "epoch": 0.9488935342031284,
      "grad_norm": 0.12155981271812312,
      "learning_rate": 1.3669051363640827e-06,
      "loss": 1.0531,
      "step": 13164
    },
    {
      "epoch": 0.9489656166654653,
      "grad_norm": 0.13402188465566,
      "learning_rate": 1.363060765239954e-06,
      "loss": 0.9763,
      "step": 13165
    },
    {
      "epoch": 0.9490376991278022,
      "grad_norm": 0.1352831945414327,
      "learning_rate": 1.3592217707114385e-06,
      "loss": 1.004,
      "step": 13166
    },
    {
      "epoch": 0.9491097815901391,
      "grad_norm": 0.1196667551144846,
      "learning_rate": 1.3553881529878132e-06,
      "loss": 0.9404,
      "step": 13167
    },
    {
      "epoch": 0.9491818640524761,
      "grad_norm": 0.12617375749839949,
      "learning_rate": 1.351559912278022e-06,
      "loss": 1.0058,
      "step": 13168
    },
    {
      "epoch": 0.949253946514813,
      "grad_norm": 0.13078478545735422,
      "learning_rate": 1.3477370487907537e-06,
      "loss": 0.9887,
      "step": 13169
    },
    {
      "epoch": 0.9493260289771499,
      "grad_norm": 0.12210430543632822,
      "learning_rate": 1.343919562734397e-06,
      "loss": 1.0469,
      "step": 13170
    },
    {
      "epoch": 0.9493981114394868,
      "grad_norm": 0.13161148228845504,
      "learning_rate": 1.3401074543170078e-06,
      "loss": 1.0259,
      "step": 13171
    },
    {
      "epoch": 0.9494701939018236,
      "grad_norm": 0.12116632093229804,
      "learning_rate": 1.3363007237464086e-06,
      "loss": 1.0129,
      "step": 13172
    },
    {
      "epoch": 0.9495422763641606,
      "grad_norm": 0.11795262801323564,
      "learning_rate": 1.3324993712300892e-06,
      "loss": 1.0944,
      "step": 13173
    },
    {
      "epoch": 0.9496143588264975,
      "grad_norm": 0.13203372397715427,
      "learning_rate": 1.3287033969752616e-06,
      "loss": 1.0031,
      "step": 13174
    },
    {
      "epoch": 0.9496864412888344,
      "grad_norm": 0.1582700406901725,
      "learning_rate": 1.324912801188838e-06,
      "loss": 1.084,
      "step": 13175
    },
    {
      "epoch": 0.9497585237511713,
      "grad_norm": 0.1299606024574438,
      "learning_rate": 1.3211275840774307e-06,
      "loss": 1.0592,
      "step": 13176
    },
    {
      "epoch": 0.9498306062135082,
      "grad_norm": 0.10764848034569212,
      "learning_rate": 1.317347745847386e-06,
      "loss": 0.9945,
      "step": 13177
    },
    {
      "epoch": 0.9499026886758452,
      "grad_norm": 0.12465636962093377,
      "learning_rate": 1.313573286704728e-06,
      "loss": 0.9891,
      "step": 13178
    },
    {
      "epoch": 0.9499747711381821,
      "grad_norm": 0.10652838456081568,
      "learning_rate": 1.309804206855203e-06,
      "loss": 1.0504,
      "step": 13179
    },
    {
      "epoch": 0.950046853600519,
      "grad_norm": 0.12304218065185174,
      "learning_rate": 1.3060405065042358e-06,
      "loss": 1.0379,
      "step": 13180
    },
    {
      "epoch": 0.9501189360628559,
      "grad_norm": 0.129057780768678,
      "learning_rate": 1.3022821858570178e-06,
      "loss": 1.0546,
      "step": 13181
    },
    {
      "epoch": 0.9501910185251928,
      "grad_norm": 0.1728699843039984,
      "learning_rate": 1.2985292451183962e-06,
      "loss": 1.0213,
      "step": 13182
    },
    {
      "epoch": 0.9502631009875298,
      "grad_norm": 0.11722764037113287,
      "learning_rate": 1.2947816844929517e-06,
      "loss": 1.0339,
      "step": 13183
    },
    {
      "epoch": 0.9503351834498667,
      "grad_norm": 0.1134488499503195,
      "learning_rate": 1.291039504184921e-06,
      "loss": 0.8935,
      "step": 13184
    },
    {
      "epoch": 0.9504072659122036,
      "grad_norm": 0.11894609094733324,
      "learning_rate": 1.28730270439833e-06,
      "loss": 1.0934,
      "step": 13185
    },
    {
      "epoch": 0.9504793483745405,
      "grad_norm": 0.13944543052278283,
      "learning_rate": 1.2835712853368487e-06,
      "loss": 1.0501,
      "step": 13186
    },
    {
      "epoch": 0.9505514308368774,
      "grad_norm": 0.11735231641098981,
      "learning_rate": 1.2798452472038814e-06,
      "loss": 1.01,
      "step": 13187
    },
    {
      "epoch": 0.9506235132992144,
      "grad_norm": 0.12847979897219827,
      "learning_rate": 1.27612459020251e-06,
      "loss": 1.005,
      "step": 13188
    },
    {
      "epoch": 0.9506955957615512,
      "grad_norm": 0.12597800132670461,
      "learning_rate": 1.272409314535561e-06,
      "loss": 1.0354,
      "step": 13189
    },
    {
      "epoch": 0.9507676782238881,
      "grad_norm": 0.16967892374096707,
      "learning_rate": 1.2686994204055503e-06,
      "loss": 1.0525,
      "step": 13190
    },
    {
      "epoch": 0.950839760686225,
      "grad_norm": 0.14636234853457625,
      "learning_rate": 1.2649949080146938e-06,
      "loss": 1.1213,
      "step": 13191
    },
    {
      "epoch": 0.9509118431485619,
      "grad_norm": 0.14069468340149513,
      "learning_rate": 1.2612957775649302e-06,
      "loss": 0.9726,
      "step": 13192
    },
    {
      "epoch": 0.9509839256108988,
      "grad_norm": 0.12367341411090027,
      "learning_rate": 1.2576020292578649e-06,
      "loss": 1.0354,
      "step": 13193
    },
    {
      "epoch": 0.9510560080732358,
      "grad_norm": 0.11445409600559663,
      "learning_rate": 1.2539136632948923e-06,
      "loss": 1.0091,
      "step": 13194
    },
    {
      "epoch": 0.9511280905355727,
      "grad_norm": 0.13768204589991215,
      "learning_rate": 1.250230679877018e-06,
      "loss": 1.0865,
      "step": 13195
    },
    {
      "epoch": 0.9512001729979096,
      "grad_norm": 0.1356716774728656,
      "learning_rate": 1.2465530792050262e-06,
      "loss": 1.0189,
      "step": 13196
    },
    {
      "epoch": 0.9512722554602465,
      "grad_norm": 0.1580639168730563,
      "learning_rate": 1.2428808614793452e-06,
      "loss": 1.0782,
      "step": 13197
    },
    {
      "epoch": 0.9513443379225834,
      "grad_norm": 0.13695590772310226,
      "learning_rate": 1.2392140269001817e-06,
      "loss": 1.0936,
      "step": 13198
    },
    {
      "epoch": 0.9514164203849204,
      "grad_norm": 0.11250585842894391,
      "learning_rate": 1.2355525756673869e-06,
      "loss": 1.0278,
      "step": 13199
    },
    {
      "epoch": 0.9514885028472573,
      "grad_norm": 0.4157171035430874,
      "learning_rate": 1.2318965079805456e-06,
      "loss": 1.037,
      "step": 13200
    },
    {
      "epoch": 0.9515605853095942,
      "grad_norm": 0.11690235335156379,
      "learning_rate": 1.2282458240389427e-06,
      "loss": 0.9609,
      "step": 13201
    },
    {
      "epoch": 0.9516326677719311,
      "grad_norm": 0.1221125227362117,
      "learning_rate": 1.2246005240415858e-06,
      "loss": 0.9353,
      "step": 13202
    },
    {
      "epoch": 0.951704750234268,
      "grad_norm": 0.1177206054967633,
      "learning_rate": 1.2209606081871717e-06,
      "loss": 1.0467,
      "step": 13203
    },
    {
      "epoch": 0.951776832696605,
      "grad_norm": 0.1167130837491268,
      "learning_rate": 1.217326076674108e-06,
      "loss": 0.9741,
      "step": 13204
    },
    {
      "epoch": 0.9518489151589419,
      "grad_norm": 0.14065629009708214,
      "learning_rate": 1.2136969297005141e-06,
      "loss": 1.0883,
      "step": 13205
    },
    {
      "epoch": 0.9519209976212788,
      "grad_norm": 0.12990296470039672,
      "learning_rate": 1.2100731674641984e-06,
      "loss": 0.9744,
      "step": 13206
    },
    {
      "epoch": 0.9519930800836156,
      "grad_norm": 0.11824622935427091,
      "learning_rate": 1.206454790162692e-06,
      "loss": 1.0419,
      "step": 13207
    },
    {
      "epoch": 0.9520651625459525,
      "grad_norm": 0.12523884015446343,
      "learning_rate": 1.2028417979932372e-06,
      "loss": 1.0811,
      "step": 13208
    },
    {
      "epoch": 0.9521372450082894,
      "grad_norm": 0.1252264875361872,
      "learning_rate": 1.1992341911527649e-06,
      "loss": 1.0865,
      "step": 13209
    },
    {
      "epoch": 0.9522093274706264,
      "grad_norm": 0.11982445065404089,
      "learning_rate": 1.1956319698379293e-06,
      "loss": 1.0079,
      "step": 13210
    },
    {
      "epoch": 0.9522814099329633,
      "grad_norm": 0.12048152809141115,
      "learning_rate": 1.192035134245073e-06,
      "loss": 1.0819,
      "step": 13211
    },
    {
      "epoch": 0.9523534923953002,
      "grad_norm": 0.14069417653965963,
      "learning_rate": 1.188443684570273e-06,
      "loss": 1.0179,
      "step": 13212
    },
    {
      "epoch": 0.9524255748576371,
      "grad_norm": 0.11675364086470301,
      "learning_rate": 1.1848576210092833e-06,
      "loss": 0.9571,
      "step": 13213
    },
    {
      "epoch": 0.952497657319974,
      "grad_norm": 0.1459089748693667,
      "learning_rate": 1.1812769437575811e-06,
      "loss": 1.0445,
      "step": 13214
    },
    {
      "epoch": 0.952569739782311,
      "grad_norm": 0.13713370844981412,
      "learning_rate": 1.1777016530103325e-06,
      "loss": 1.0176,
      "step": 13215
    },
    {
      "epoch": 0.9526418222446479,
      "grad_norm": 0.12288140672964613,
      "learning_rate": 1.174131748962437e-06,
      "loss": 0.9911,
      "step": 13216
    },
    {
      "epoch": 0.9527139047069848,
      "grad_norm": 0.12662284171693594,
      "learning_rate": 1.1705672318084947e-06,
      "loss": 0.9788,
      "step": 13217
    },
    {
      "epoch": 0.9527859871693217,
      "grad_norm": 0.1089281311616668,
      "learning_rate": 1.1670081017427836e-06,
      "loss": 0.9829,
      "step": 13218
    },
    {
      "epoch": 0.9528580696316586,
      "grad_norm": 0.13655871789886545,
      "learning_rate": 1.163454358959315e-06,
      "loss": 1.0535,
      "step": 13219
    },
    {
      "epoch": 0.9529301520939956,
      "grad_norm": 0.21746824226191228,
      "learning_rate": 1.1599060036518005e-06,
      "loss": 1.02,
      "step": 13220
    },
    {
      "epoch": 0.9530022345563325,
      "grad_norm": 0.16343953048202906,
      "learning_rate": 1.1563630360136523e-06,
      "loss": 0.9197,
      "step": 13221
    },
    {
      "epoch": 0.9530743170186694,
      "grad_norm": 0.1654953556319771,
      "learning_rate": 1.1528254562380048e-06,
      "loss": 0.9867,
      "step": 13222
    },
    {
      "epoch": 0.9531463994810063,
      "grad_norm": 0.11427930824454514,
      "learning_rate": 1.1492932645176924e-06,
      "loss": 1.0432,
      "step": 13223
    },
    {
      "epoch": 0.9532184819433431,
      "grad_norm": 0.1174189051820554,
      "learning_rate": 1.1457664610452168e-06,
      "loss": 1.0201,
      "step": 13224
    },
    {
      "epoch": 0.95329056440568,
      "grad_norm": 0.10788158581870154,
      "learning_rate": 1.1422450460128687e-06,
      "loss": 1.0648,
      "step": 13225
    },
    {
      "epoch": 0.953362646868017,
      "grad_norm": 0.1101330986022403,
      "learning_rate": 1.138729019612561e-06,
      "loss": 1.0793,
      "step": 13226
    },
    {
      "epoch": 0.9534347293303539,
      "grad_norm": 0.12294833831389605,
      "learning_rate": 1.135218382035963e-06,
      "loss": 1.1048,
      "step": 13227
    },
    {
      "epoch": 0.9535068117926908,
      "grad_norm": 0.33161864330533697,
      "learning_rate": 1.1317131334744212e-06,
      "loss": 1.0622,
      "step": 13228
    },
    {
      "epoch": 0.9535788942550277,
      "grad_norm": 0.11322613069632394,
      "learning_rate": 1.1282132741190278e-06,
      "loss": 0.9989,
      "step": 13229
    },
    {
      "epoch": 0.9536509767173647,
      "grad_norm": 0.13174742664906364,
      "learning_rate": 1.1247188041605405e-06,
      "loss": 0.9967,
      "step": 13230
    },
    {
      "epoch": 0.9537230591797016,
      "grad_norm": 0.12978239413897757,
      "learning_rate": 1.121229723789441e-06,
      "loss": 1.0122,
      "step": 13231
    },
    {
      "epoch": 0.9537951416420385,
      "grad_norm": 0.12012560875678255,
      "learning_rate": 1.1177460331959212e-06,
      "loss": 0.9898,
      "step": 13232
    },
    {
      "epoch": 0.9538672241043754,
      "grad_norm": 0.13279394486527787,
      "learning_rate": 1.1142677325698515e-06,
      "loss": 1.0073,
      "step": 13233
    },
    {
      "epoch": 0.9539393065667123,
      "grad_norm": 0.14810671879747017,
      "learning_rate": 1.110794822100858e-06,
      "loss": 1.0356,
      "step": 13234
    },
    {
      "epoch": 0.9540113890290493,
      "grad_norm": 0.12968089836664165,
      "learning_rate": 1.107327301978245e-06,
      "loss": 1.0668,
      "step": 13235
    },
    {
      "epoch": 0.9540834714913862,
      "grad_norm": 0.17395414227247838,
      "learning_rate": 1.1038651723910055e-06,
      "loss": 0.9877,
      "step": 13236
    },
    {
      "epoch": 0.9541555539537231,
      "grad_norm": 0.12004544938113117,
      "learning_rate": 1.1004084335278553e-06,
      "loss": 1.0487,
      "step": 13237
    },
    {
      "epoch": 0.95422763641606,
      "grad_norm": 0.12695700321772208,
      "learning_rate": 1.0969570855772327e-06,
      "loss": 1.0141,
      "step": 13238
    },
    {
      "epoch": 0.9542997188783969,
      "grad_norm": 0.1594164186801164,
      "learning_rate": 1.093511128727276e-06,
      "loss": 0.9762,
      "step": 13239
    },
    {
      "epoch": 0.9543718013407338,
      "grad_norm": 0.12720393790719647,
      "learning_rate": 1.0900705631657904e-06,
      "loss": 1.0229,
      "step": 13240
    },
    {
      "epoch": 0.9544438838030707,
      "grad_norm": 0.18084378257294417,
      "learning_rate": 1.086635389080326e-06,
      "loss": 1.0672,
      "step": 13241
    },
    {
      "epoch": 0.9545159662654076,
      "grad_norm": 0.11528589785684658,
      "learning_rate": 1.0832056066581442e-06,
      "loss": 1.0394,
      "step": 13242
    },
    {
      "epoch": 0.9545880487277445,
      "grad_norm": 0.11739876571350677,
      "learning_rate": 1.0797812160861954e-06,
      "loss": 1.0456,
      "step": 13243
    },
    {
      "epoch": 0.9546601311900814,
      "grad_norm": 0.12070100683199053,
      "learning_rate": 1.0763622175511413e-06,
      "loss": 1.023,
      "step": 13244
    },
    {
      "epoch": 0.9547322136524183,
      "grad_norm": 0.11449100511446361,
      "learning_rate": 1.0729486112393327e-06,
      "loss": 1.0442,
      "step": 13245
    },
    {
      "epoch": 0.9548042961147553,
      "grad_norm": 0.12800208415196654,
      "learning_rate": 1.0695403973368434e-06,
      "loss": 1.0513,
      "step": 13246
    },
    {
      "epoch": 0.9548763785770922,
      "grad_norm": 0.10823101306791891,
      "learning_rate": 1.0661375760294801e-06,
      "loss": 1.0435,
      "step": 13247
    },
    {
      "epoch": 0.9549484610394291,
      "grad_norm": 0.13120250029611238,
      "learning_rate": 1.0627401475026944e-06,
      "loss": 1.1168,
      "step": 13248
    },
    {
      "epoch": 0.955020543501766,
      "grad_norm": 0.11057432733910211,
      "learning_rate": 1.0593481119416826e-06,
      "loss": 1.0375,
      "step": 13249
    },
    {
      "epoch": 0.9550926259641029,
      "grad_norm": 0.12733468047565003,
      "learning_rate": 1.0559614695313525e-06,
      "loss": 1.0824,
      "step": 13250
    },
    {
      "epoch": 0.9551647084264399,
      "grad_norm": 0.2016237362818377,
      "learning_rate": 1.0525802204563118e-06,
      "loss": 1.0753,
      "step": 13251
    },
    {
      "epoch": 0.9552367908887768,
      "grad_norm": 0.12594000031064875,
      "learning_rate": 1.0492043649008464e-06,
      "loss": 1.016,
      "step": 13252
    },
    {
      "epoch": 0.9553088733511137,
      "grad_norm": 0.15109234991648154,
      "learning_rate": 1.0458339030489872e-06,
      "loss": 1.0729,
      "step": 13253
    },
    {
      "epoch": 0.9553809558134506,
      "grad_norm": 0.10977078667203283,
      "learning_rate": 1.0424688350844425e-06,
      "loss": 1.0477,
      "step": 13254
    },
    {
      "epoch": 0.9554530382757875,
      "grad_norm": 0.3603983907294268,
      "learning_rate": 1.0391091611906544e-06,
      "loss": 1.1149,
      "step": 13255
    },
    {
      "epoch": 0.9555251207381245,
      "grad_norm": 0.12475691295427462,
      "learning_rate": 1.0357548815507434e-06,
      "loss": 1.0013,
      "step": 13256
    },
    {
      "epoch": 0.9555972032004614,
      "grad_norm": 0.1276763429627298,
      "learning_rate": 1.0324059963475518e-06,
      "loss": 0.921,
      "step": 13257
    },
    {
      "epoch": 0.9556692856627983,
      "grad_norm": 0.18577324418517344,
      "learning_rate": 1.0290625057636227e-06,
      "loss": 1.0143,
      "step": 13258
    },
    {
      "epoch": 0.9557413681251351,
      "grad_norm": 0.11944407739226119,
      "learning_rate": 1.025724409981199e-06,
      "loss": 0.9471,
      "step": 13259
    },
    {
      "epoch": 0.955813450587472,
      "grad_norm": 0.25472642880021634,
      "learning_rate": 1.0223917091822466e-06,
      "loss": 0.996,
      "step": 13260
    },
    {
      "epoch": 0.9558855330498089,
      "grad_norm": 0.11584365135944283,
      "learning_rate": 1.0190644035484308e-06,
      "loss": 1.0478,
      "step": 13261
    },
    {
      "epoch": 0.9559576155121459,
      "grad_norm": 0.11368336841008661,
      "learning_rate": 1.015742493261107e-06,
      "loss": 1.1078,
      "step": 13262
    },
    {
      "epoch": 0.9560296979744828,
      "grad_norm": 0.12836548231321793,
      "learning_rate": 1.0124259785013635e-06,
      "loss": 1.0482,
      "step": 13263
    },
    {
      "epoch": 0.9561017804368197,
      "grad_norm": 0.14389880633027516,
      "learning_rate": 1.0091148594499666e-06,
      "loss": 1.0308,
      "step": 13264
    },
    {
      "epoch": 0.9561738628991566,
      "grad_norm": 0.11785141819214145,
      "learning_rate": 1.0058091362874168e-06,
      "loss": 1.0419,
      "step": 13265
    },
    {
      "epoch": 0.9562459453614935,
      "grad_norm": 0.12739814277044376,
      "learning_rate": 1.0025088091938806e-06,
      "loss": 1.0149,
      "step": 13266
    },
    {
      "epoch": 0.9563180278238305,
      "grad_norm": 0.11830066003962716,
      "learning_rate": 9.99213878349281e-07,
      "loss": 1.0199,
      "step": 13267
    },
    {
      "epoch": 0.9563901102861674,
      "grad_norm": 0.18288174338748142,
      "learning_rate": 9.95924343933219e-07,
      "loss": 1.0129,
      "step": 13268
    },
    {
      "epoch": 0.9564621927485043,
      "grad_norm": 0.18210784073522693,
      "learning_rate": 9.92640206124984e-07,
      "loss": 1.0177,
      "step": 13269
    },
    {
      "epoch": 0.9565342752108412,
      "grad_norm": 0.1230454808552929,
      "learning_rate": 9.893614651035998e-07,
      "loss": 1.0842,
      "step": 13270
    },
    {
      "epoch": 0.9566063576731781,
      "grad_norm": 0.14004942904564796,
      "learning_rate": 9.860881210478012e-07,
      "loss": 1.0095,
      "step": 13271
    },
    {
      "epoch": 0.9566784401355151,
      "grad_norm": 0.11495991734233464,
      "learning_rate": 9.82820174136001e-07,
      "loss": 1.0977,
      "step": 13272
    },
    {
      "epoch": 0.956750522597852,
      "grad_norm": 0.1268613973087549,
      "learning_rate": 9.795576245463345e-07,
      "loss": 0.9977,
      "step": 13273
    },
    {
      "epoch": 0.9568226050601889,
      "grad_norm": 0.11072546933711953,
      "learning_rate": 9.76300472456637e-07,
      "loss": 1.028,
      "step": 13274
    },
    {
      "epoch": 0.9568946875225258,
      "grad_norm": 0.12293867687189224,
      "learning_rate": 9.730487180444558e-07,
      "loss": 0.9874,
      "step": 13275
    },
    {
      "epoch": 0.9569667699848626,
      "grad_norm": 0.12808296936948457,
      "learning_rate": 9.698023614870488e-07,
      "loss": 1.0394,
      "step": 13276
    },
    {
      "epoch": 0.9570388524471996,
      "grad_norm": 0.12651134430000377,
      "learning_rate": 9.665614029613413e-07,
      "loss": 1.0313,
      "step": 13277
    },
    {
      "epoch": 0.9571109349095365,
      "grad_norm": 0.1508862192298974,
      "learning_rate": 9.63325842644025e-07,
      "loss": 0.9555,
      "step": 13278
    },
    {
      "epoch": 0.9571830173718734,
      "grad_norm": 0.11403152218822907,
      "learning_rate": 9.600956807114703e-07,
      "loss": 1.0498,
      "step": 13279
    },
    {
      "epoch": 0.9572550998342103,
      "grad_norm": 0.11967195933671329,
      "learning_rate": 9.56870917339725e-07,
      "loss": 0.8637,
      "step": 13280
    },
    {
      "epoch": 0.9573271822965472,
      "grad_norm": 0.1276924220784229,
      "learning_rate": 9.53651552704582e-07,
      "loss": 1.0723,
      "step": 13281
    },
    {
      "epoch": 0.9573992647588841,
      "grad_norm": 0.12302813430580976,
      "learning_rate": 9.504375869815229e-07,
      "loss": 0.9684,
      "step": 13282
    },
    {
      "epoch": 0.9574713472212211,
      "grad_norm": 0.11126269652223175,
      "learning_rate": 9.472290203457412e-07,
      "loss": 1.0558,
      "step": 13283
    },
    {
      "epoch": 0.957543429683558,
      "grad_norm": 0.17248251586160274,
      "learning_rate": 9.440258529721302e-07,
      "loss": 0.999,
      "step": 13284
    },
    {
      "epoch": 0.9576155121458949,
      "grad_norm": 0.11242334016332647,
      "learning_rate": 9.408280850352836e-07,
      "loss": 1.0604,
      "step": 13285
    },
    {
      "epoch": 0.9576875946082318,
      "grad_norm": 0.11481684306613267,
      "learning_rate": 9.376357167095174e-07,
      "loss": 1.0283,
      "step": 13286
    },
    {
      "epoch": 0.9577596770705687,
      "grad_norm": 0.13488551265385934,
      "learning_rate": 9.34448748168848e-07,
      "loss": 0.9967,
      "step": 13287
    },
    {
      "epoch": 0.9578317595329057,
      "grad_norm": 0.12696430847079296,
      "learning_rate": 9.31267179586981e-07,
      "loss": 0.9721,
      "step": 13288
    },
    {
      "epoch": 0.9579038419952426,
      "grad_norm": 0.11263615815584074,
      "learning_rate": 9.280910111373553e-07,
      "loss": 1.0832,
      "step": 13289
    },
    {
      "epoch": 0.9579759244575795,
      "grad_norm": 0.12627589067458392,
      "learning_rate": 9.24920242993077e-07,
      "loss": 0.9735,
      "step": 13290
    },
    {
      "epoch": 0.9580480069199164,
      "grad_norm": 0.15771975613594416,
      "learning_rate": 9.217548753270078e-07,
      "loss": 1.0789,
      "step": 13291
    },
    {
      "epoch": 0.9581200893822533,
      "grad_norm": 0.12479909071178723,
      "learning_rate": 9.185949083116763e-07,
      "loss": 1.1601,
      "step": 13292
    },
    {
      "epoch": 0.9581921718445902,
      "grad_norm": 0.1254375488148498,
      "learning_rate": 9.154403421193225e-07,
      "loss": 0.9814,
      "step": 13293
    },
    {
      "epoch": 0.9582642543069271,
      "grad_norm": 0.1098607005536254,
      "learning_rate": 9.122911769219089e-07,
      "loss": 1.0226,
      "step": 13294
    },
    {
      "epoch": 0.958336336769264,
      "grad_norm": 0.2841858210990875,
      "learning_rate": 9.091474128910982e-07,
      "loss": 1.0504,
      "step": 13295
    },
    {
      "epoch": 0.9584084192316009,
      "grad_norm": 0.12384030241161374,
      "learning_rate": 9.060090501982421e-07,
      "loss": 1.0381,
      "step": 13296
    },
    {
      "epoch": 0.9584805016939378,
      "grad_norm": 0.12309006891806056,
      "learning_rate": 9.02876089014415e-07,
      "loss": 1.024,
      "step": 13297
    },
    {
      "epoch": 0.9585525841562748,
      "grad_norm": 0.11693951327251173,
      "learning_rate": 8.997485295103802e-07,
      "loss": 1.0636,
      "step": 13298
    },
    {
      "epoch": 0.9586246666186117,
      "grad_norm": 0.13180840149314266,
      "learning_rate": 8.966263718566237e-07,
      "loss": 1.0024,
      "step": 13299
    },
    {
      "epoch": 0.9586967490809486,
      "grad_norm": 0.1062490795851812,
      "learning_rate": 8.935096162233536e-07,
      "loss": 0.9613,
      "step": 13300
    },
    {
      "epoch": 0.9587688315432855,
      "grad_norm": 0.11264128588851564,
      "learning_rate": 8.90398262780423e-07,
      "loss": 1.0768,
      "step": 13301
    },
    {
      "epoch": 0.9588409140056224,
      "grad_norm": 0.12753733653291846,
      "learning_rate": 8.872923116974518e-07,
      "loss": 1.056,
      "step": 13302
    },
    {
      "epoch": 0.9589129964679594,
      "grad_norm": 0.14295905452642937,
      "learning_rate": 8.84191763143738e-07,
      "loss": 1.0411,
      "step": 13303
    },
    {
      "epoch": 0.9589850789302963,
      "grad_norm": 0.14795375792806914,
      "learning_rate": 8.810966172882907e-07,
      "loss": 1.0139,
      "step": 13304
    },
    {
      "epoch": 0.9590571613926332,
      "grad_norm": 0.2436339706291158,
      "learning_rate": 8.780068742998193e-07,
      "loss": 1.0187,
      "step": 13305
    },
    {
      "epoch": 0.9591292438549701,
      "grad_norm": 0.11704666228558834,
      "learning_rate": 8.749225343467449e-07,
      "loss": 1.0241,
      "step": 13306
    },
    {
      "epoch": 0.959201326317307,
      "grad_norm": 0.11538469928667615,
      "learning_rate": 8.718435975971994e-07,
      "loss": 1.0543,
      "step": 13307
    },
    {
      "epoch": 0.959273408779644,
      "grad_norm": 0.11313585585497743,
      "learning_rate": 8.687700642189934e-07,
      "loss": 1.0056,
      "step": 13308
    },
    {
      "epoch": 0.9593454912419809,
      "grad_norm": 0.1721093463684057,
      "learning_rate": 8.657019343796702e-07,
      "loss": 1.0347,
      "step": 13309
    },
    {
      "epoch": 0.9594175737043178,
      "grad_norm": 0.15570951616362055,
      "learning_rate": 8.62639208246474e-07,
      "loss": 1.0194,
      "step": 13310
    },
    {
      "epoch": 0.9594896561666546,
      "grad_norm": 0.12002105372332181,
      "learning_rate": 8.595818859863491e-07,
      "loss": 0.9696,
      "step": 13311
    },
    {
      "epoch": 0.9595617386289915,
      "grad_norm": 0.11583297531209275,
      "learning_rate": 8.565299677659511e-07,
      "loss": 1.0118,
      "step": 13312
    },
    {
      "epoch": 0.9596338210913284,
      "grad_norm": 0.12221339869357961,
      "learning_rate": 8.534834537516245e-07,
      "loss": 1.0224,
      "step": 13313
    },
    {
      "epoch": 0.9597059035536654,
      "grad_norm": 0.139231743350302,
      "learning_rate": 8.504423441094478e-07,
      "loss": 1.0618,
      "step": 13314
    },
    {
      "epoch": 0.9597779860160023,
      "grad_norm": 0.11871194976314853,
      "learning_rate": 8.474066390051771e-07,
      "loss": 1.04,
      "step": 13315
    },
    {
      "epoch": 0.9598500684783392,
      "grad_norm": 0.12629167300099362,
      "learning_rate": 8.443763386042914e-07,
      "loss": 1.0684,
      "step": 13316
    },
    {
      "epoch": 0.9599221509406761,
      "grad_norm": 0.12530645675673452,
      "learning_rate": 8.413514430719694e-07,
      "loss": 1.0193,
      "step": 13317
    },
    {
      "epoch": 0.959994233403013,
      "grad_norm": 0.11834839832527891,
      "learning_rate": 8.383319525730793e-07,
      "loss": 1.0403,
      "step": 13318
    },
    {
      "epoch": 0.96006631586535,
      "grad_norm": 0.1978583592841791,
      "learning_rate": 8.353178672722339e-07,
      "loss": 1.0221,
      "step": 13319
    },
    {
      "epoch": 0.9601383983276869,
      "grad_norm": 0.12841893839900878,
      "learning_rate": 8.323091873337241e-07,
      "loss": 1.1191,
      "step": 13320
    },
    {
      "epoch": 0.9602104807900238,
      "grad_norm": 0.10839488201505944,
      "learning_rate": 8.293059129215297e-07,
      "loss": 1.0227,
      "step": 13321
    },
    {
      "epoch": 0.9602825632523607,
      "grad_norm": 0.11574631130127039,
      "learning_rate": 8.263080441993753e-07,
      "loss": 0.954,
      "step": 13322
    },
    {
      "epoch": 0.9603546457146976,
      "grad_norm": 0.1202774348428818,
      "learning_rate": 8.233155813306637e-07,
      "loss": 0.9955,
      "step": 13323
    },
    {
      "epoch": 0.9604267281770346,
      "grad_norm": 0.14005289350889222,
      "learning_rate": 8.203285244785198e-07,
      "loss": 1.007,
      "step": 13324
    },
    {
      "epoch": 0.9604988106393715,
      "grad_norm": 0.11710597089903671,
      "learning_rate": 8.173468738057466e-07,
      "loss": 1.0007,
      "step": 13325
    },
    {
      "epoch": 0.9605708931017084,
      "grad_norm": 0.11982363871987486,
      "learning_rate": 8.14370629474892e-07,
      "loss": 1.0388,
      "step": 13326
    },
    {
      "epoch": 0.9606429755640453,
      "grad_norm": 0.12771067032853825,
      "learning_rate": 8.113997916481708e-07,
      "loss": 0.8897,
      "step": 13327
    },
    {
      "epoch": 0.9607150580263821,
      "grad_norm": 0.11832669920226911,
      "learning_rate": 8.084343604875311e-07,
      "loss": 1.0204,
      "step": 13328
    },
    {
      "epoch": 0.960787140488719,
      "grad_norm": 0.12458908104609076,
      "learning_rate": 8.054743361546213e-07,
      "loss": 0.9861,
      "step": 13329
    },
    {
      "epoch": 0.960859222951056,
      "grad_norm": 0.12256998493087476,
      "learning_rate": 8.02519718810757e-07,
      "loss": 0.9929,
      "step": 13330
    },
    {
      "epoch": 0.9609313054133929,
      "grad_norm": 0.11987093688845,
      "learning_rate": 7.995705086170313e-07,
      "loss": 1.0467,
      "step": 13331
    },
    {
      "epoch": 0.9610033878757298,
      "grad_norm": 0.12128779955152437,
      "learning_rate": 7.966267057341826e-07,
      "loss": 1.0439,
      "step": 13332
    },
    {
      "epoch": 0.9610754703380667,
      "grad_norm": 0.12637551598116206,
      "learning_rate": 7.936883103226822e-07,
      "loss": 0.9795,
      "step": 13333
    },
    {
      "epoch": 0.9611475528004036,
      "grad_norm": 0.1447488298367098,
      "learning_rate": 7.907553225426801e-07,
      "loss": 1.0315,
      "step": 13334
    },
    {
      "epoch": 0.9612196352627406,
      "grad_norm": 0.11154484828452794,
      "learning_rate": 7.878277425540703e-07,
      "loss": 1.0017,
      "step": 13335
    },
    {
      "epoch": 0.9612917177250775,
      "grad_norm": 0.11688821065830508,
      "learning_rate": 7.849055705164255e-07,
      "loss": 0.9584,
      "step": 13336
    },
    {
      "epoch": 0.9613638001874144,
      "grad_norm": 0.10581847470937521,
      "learning_rate": 7.819888065890402e-07,
      "loss": 1.0267,
      "step": 13337
    },
    {
      "epoch": 0.9614358826497513,
      "grad_norm": 0.11350497820903824,
      "learning_rate": 7.790774509308873e-07,
      "loss": 0.992,
      "step": 13338
    },
    {
      "epoch": 0.9615079651120882,
      "grad_norm": 0.13662771576127064,
      "learning_rate": 7.761715037006623e-07,
      "loss": 1.0172,
      "step": 13339
    },
    {
      "epoch": 0.9615800475744252,
      "grad_norm": 0.1102761406750305,
      "learning_rate": 7.732709650567826e-07,
      "loss": 1.0401,
      "step": 13340
    },
    {
      "epoch": 0.9616521300367621,
      "grad_norm": 0.1280700735036804,
      "learning_rate": 7.703758351573332e-07,
      "loss": 1.0752,
      "step": 13341
    },
    {
      "epoch": 0.961724212499099,
      "grad_norm": 0.298532850247206,
      "learning_rate": 7.67486114160143e-07,
      "loss": 1.0988,
      "step": 13342
    },
    {
      "epoch": 0.9617962949614359,
      "grad_norm": 0.12111810734325436,
      "learning_rate": 7.646018022227086e-07,
      "loss": 1.0664,
      "step": 13343
    },
    {
      "epoch": 0.9618683774237728,
      "grad_norm": 0.165416210467467,
      "learning_rate": 7.617228995022707e-07,
      "loss": 1.0289,
      "step": 13344
    },
    {
      "epoch": 0.9619404598861097,
      "grad_norm": 0.17413999023040289,
      "learning_rate": 7.588494061557483e-07,
      "loss": 1.0486,
      "step": 13345
    },
    {
      "epoch": 0.9620125423484466,
      "grad_norm": 0.13694872593495383,
      "learning_rate": 7.559813223397605e-07,
      "loss": 1.0577,
      "step": 13346
    },
    {
      "epoch": 0.9620846248107835,
      "grad_norm": 0.11519332721210317,
      "learning_rate": 7.5311864821066e-07,
      "loss": 1.0506,
      "step": 13347
    },
    {
      "epoch": 0.9621567072731204,
      "grad_norm": 0.17971437461045306,
      "learning_rate": 7.502613839244776e-07,
      "loss": 1.0594,
      "step": 13348
    },
    {
      "epoch": 0.9622287897354573,
      "grad_norm": 0.12133734037364162,
      "learning_rate": 7.474095296369665e-07,
      "loss": 1.0438,
      "step": 13349
    },
    {
      "epoch": 0.9623008721977943,
      "grad_norm": 0.12832671597097847,
      "learning_rate": 7.44563085503569e-07,
      "loss": 1.0482,
      "step": 13350
    },
    {
      "epoch": 0.9623729546601312,
      "grad_norm": 0.1536153330264548,
      "learning_rate": 7.417220516794499e-07,
      "loss": 0.9555,
      "step": 13351
    },
    {
      "epoch": 0.9624450371224681,
      "grad_norm": 0.12533894500508877,
      "learning_rate": 7.388864283194741e-07,
      "loss": 1.047,
      "step": 13352
    },
    {
      "epoch": 0.962517119584805,
      "grad_norm": 0.13227380161741337,
      "learning_rate": 7.360562155782069e-07,
      "loss": 1.096,
      "step": 13353
    },
    {
      "epoch": 0.9625892020471419,
      "grad_norm": 0.11765379121604448,
      "learning_rate": 7.332314136099027e-07,
      "loss": 0.9964,
      "step": 13354
    },
    {
      "epoch": 0.9626612845094789,
      "grad_norm": 0.1865797452659009,
      "learning_rate": 7.304120225685607e-07,
      "loss": 1.061,
      "step": 13355
    },
    {
      "epoch": 0.9627333669718158,
      "grad_norm": 0.1333380949148248,
      "learning_rate": 7.275980426078577e-07,
      "loss": 1.0257,
      "step": 13356
    },
    {
      "epoch": 0.9628054494341527,
      "grad_norm": 0.11609279012363671,
      "learning_rate": 7.247894738811711e-07,
      "loss": 1.0807,
      "step": 13357
    },
    {
      "epoch": 0.9628775318964896,
      "grad_norm": 0.13967951785233537,
      "learning_rate": 7.219863165416008e-07,
      "loss": 1.0558,
      "step": 13358
    },
    {
      "epoch": 0.9629496143588265,
      "grad_norm": 0.11697300169346649,
      "learning_rate": 7.191885707419465e-07,
      "loss": 1.0496,
      "step": 13359
    },
    {
      "epoch": 0.9630216968211635,
      "grad_norm": 0.17797488755629262,
      "learning_rate": 7.163962366346977e-07,
      "loss": 1.0693,
      "step": 13360
    },
    {
      "epoch": 0.9630937792835004,
      "grad_norm": 0.13336476253558513,
      "learning_rate": 7.136093143720767e-07,
      "loss": 1.0443,
      "step": 13361
    },
    {
      "epoch": 0.9631658617458373,
      "grad_norm": 0.2777521379371853,
      "learning_rate": 7.108278041059957e-07,
      "loss": 1.1044,
      "step": 13362
    },
    {
      "epoch": 0.9632379442081741,
      "grad_norm": 0.11849730829933451,
      "learning_rate": 7.080517059880553e-07,
      "loss": 1.047,
      "step": 13363
    },
    {
      "epoch": 0.963310026670511,
      "grad_norm": 0.13495340852369833,
      "learning_rate": 7.052810201695903e-07,
      "loss": 1.0331,
      "step": 13364
    },
    {
      "epoch": 0.9633821091328479,
      "grad_norm": 0.2060463474464177,
      "learning_rate": 7.025157468016241e-07,
      "loss": 1.0128,
      "step": 13365
    },
    {
      "epoch": 0.9634541915951849,
      "grad_norm": 0.27282941477488587,
      "learning_rate": 6.99755886034903e-07,
      "loss": 1.0424,
      "step": 13366
    },
    {
      "epoch": 0.9635262740575218,
      "grad_norm": 0.12395439808097067,
      "learning_rate": 6.970014380198397e-07,
      "loss": 1.0221,
      "step": 13367
    },
    {
      "epoch": 0.9635983565198587,
      "grad_norm": 0.13201946262244083,
      "learning_rate": 6.942524029065922e-07,
      "loss": 1.0817,
      "step": 13368
    },
    {
      "epoch": 0.9636704389821956,
      "grad_norm": 0.14056732988179615,
      "learning_rate": 6.91508780844996e-07,
      "loss": 1.0226,
      "step": 13369
    },
    {
      "epoch": 0.9637425214445325,
      "grad_norm": 0.17603251165633105,
      "learning_rate": 6.887705719846205e-07,
      "loss": 1.0142,
      "step": 13370
    },
    {
      "epoch": 0.9638146039068695,
      "grad_norm": 0.11770817246759828,
      "learning_rate": 6.860377764747017e-07,
      "loss": 0.959,
      "step": 13371
    },
    {
      "epoch": 0.9638866863692064,
      "grad_norm": 0.15463858574158854,
      "learning_rate": 6.833103944642205e-07,
      "loss": 0.9738,
      "step": 13372
    },
    {
      "epoch": 0.9639587688315433,
      "grad_norm": 0.14463710539962427,
      "learning_rate": 6.805884261018358e-07,
      "loss": 1.024,
      "step": 13373
    },
    {
      "epoch": 0.9640308512938802,
      "grad_norm": 0.12141247681491453,
      "learning_rate": 6.778718715359178e-07,
      "loss": 1.0384,
      "step": 13374
    },
    {
      "epoch": 0.9641029337562171,
      "grad_norm": 0.1076218399258925,
      "learning_rate": 6.751607309145369e-07,
      "loss": 1.0242,
      "step": 13375
    },
    {
      "epoch": 0.9641750162185541,
      "grad_norm": 0.12325439592551578,
      "learning_rate": 6.72455004385486e-07,
      "loss": 1.0336,
      "step": 13376
    },
    {
      "epoch": 0.964247098680891,
      "grad_norm": 0.15836841901762813,
      "learning_rate": 6.697546920962583e-07,
      "loss": 1.0123,
      "step": 13377
    },
    {
      "epoch": 0.9643191811432279,
      "grad_norm": 0.13134288431719368,
      "learning_rate": 6.670597941940137e-07,
      "loss": 1.0395,
      "step": 13378
    },
    {
      "epoch": 0.9643912636055648,
      "grad_norm": 0.12312570366791786,
      "learning_rate": 6.643703108256794e-07,
      "loss": 1.018,
      "step": 13379
    },
    {
      "epoch": 0.9644633460679016,
      "grad_norm": 0.13888261651458955,
      "learning_rate": 6.616862421378489e-07,
      "loss": 0.9749,
      "step": 13380
    },
    {
      "epoch": 0.9645354285302385,
      "grad_norm": 0.16528404871427696,
      "learning_rate": 6.590075882768166e-07,
      "loss": 1.0215,
      "step": 13381
    },
    {
      "epoch": 0.9646075109925755,
      "grad_norm": 0.12421284707839647,
      "learning_rate": 6.563343493886099e-07,
      "loss": 1.0143,
      "step": 13382
    },
    {
      "epoch": 0.9646795934549124,
      "grad_norm": 0.1358832677796433,
      "learning_rate": 6.536665256189234e-07,
      "loss": 1.0649,
      "step": 13383
    },
    {
      "epoch": 0.9647516759172493,
      "grad_norm": 0.11939560458999278,
      "learning_rate": 6.510041171131853e-07,
      "loss": 1.0253,
      "step": 13384
    },
    {
      "epoch": 0.9648237583795862,
      "grad_norm": 0.1328685572741252,
      "learning_rate": 6.483471240165462e-07,
      "loss": 1.0614,
      "step": 13385
    },
    {
      "epoch": 0.9648958408419231,
      "grad_norm": 0.1408188732658539,
      "learning_rate": 6.456955464737902e-07,
      "loss": 1.0794,
      "step": 13386
    },
    {
      "epoch": 0.9649679233042601,
      "grad_norm": 0.11589248547920115,
      "learning_rate": 6.430493846294904e-07,
      "loss": 1.0084,
      "step": 13387
    },
    {
      "epoch": 0.965040005766597,
      "grad_norm": 0.12074523095794819,
      "learning_rate": 6.404086386278651e-07,
      "loss": 1.0669,
      "step": 13388
    },
    {
      "epoch": 0.9651120882289339,
      "grad_norm": 0.12550212287401083,
      "learning_rate": 6.377733086128767e-07,
      "loss": 0.995,
      "step": 13389
    },
    {
      "epoch": 0.9651841706912708,
      "grad_norm": 0.12482184576921371,
      "learning_rate": 6.351433947281438e-07,
      "loss": 1.0317,
      "step": 13390
    },
    {
      "epoch": 0.9652562531536077,
      "grad_norm": 0.12676796796972828,
      "learning_rate": 6.325188971170515e-07,
      "loss": 0.9491,
      "step": 13391
    },
    {
      "epoch": 0.9653283356159447,
      "grad_norm": 0.15072407205962052,
      "learning_rate": 6.29899815922641e-07,
      "loss": 1.1779,
      "step": 13392
    },
    {
      "epoch": 0.9654004180782816,
      "grad_norm": 0.13083004629107103,
      "learning_rate": 6.272861512876871e-07,
      "loss": 1.0475,
      "step": 13393
    },
    {
      "epoch": 0.9654725005406185,
      "grad_norm": 0.11401969819273325,
      "learning_rate": 6.246779033546535e-07,
      "loss": 1.0818,
      "step": 13394
    },
    {
      "epoch": 0.9655445830029554,
      "grad_norm": 0.10730529446312254,
      "learning_rate": 6.220750722657043e-07,
      "loss": 0.944,
      "step": 13395
    },
    {
      "epoch": 0.9656166654652923,
      "grad_norm": 0.14066900439758706,
      "learning_rate": 6.19477658162726e-07,
      "loss": 0.9989,
      "step": 13396
    },
    {
      "epoch": 0.9656887479276292,
      "grad_norm": 0.12306808965911646,
      "learning_rate": 6.168856611872942e-07,
      "loss": 1.0248,
      "step": 13397
    },
    {
      "epoch": 0.9657608303899661,
      "grad_norm": 0.125369287726494,
      "learning_rate": 6.14299081480707e-07,
      "loss": 0.9931,
      "step": 13398
    },
    {
      "epoch": 0.965832912852303,
      "grad_norm": 0.13853582109763143,
      "learning_rate": 6.117179191839517e-07,
      "loss": 0.974,
      "step": 13399
    },
    {
      "epoch": 0.9659049953146399,
      "grad_norm": 0.12486093079318122,
      "learning_rate": 6.091421744377157e-07,
      "loss": 1.0285,
      "step": 13400
    },
    {
      "epoch": 0.9659770777769768,
      "grad_norm": 0.11642347676212678,
      "learning_rate": 6.065718473824089e-07,
      "loss": 1.0223,
      "step": 13401
    },
    {
      "epoch": 0.9660491602393138,
      "grad_norm": 0.12302809802456031,
      "learning_rate": 6.040069381581415e-07,
      "loss": 0.9449,
      "step": 13402
    },
    {
      "epoch": 0.9661212427016507,
      "grad_norm": 0.15702584076876486,
      "learning_rate": 6.014474469047126e-07,
      "loss": 1.0073,
      "step": 13403
    },
    {
      "epoch": 0.9661933251639876,
      "grad_norm": 0.1250829941697826,
      "learning_rate": 5.988933737616442e-07,
      "loss": 0.9772,
      "step": 13404
    },
    {
      "epoch": 0.9662654076263245,
      "grad_norm": 0.13525632611788566,
      "learning_rate": 5.963447188681581e-07,
      "loss": 1.0167,
      "step": 13405
    },
    {
      "epoch": 0.9663374900886614,
      "grad_norm": 0.18322840719263242,
      "learning_rate": 5.938014823631655e-07,
      "loss": 1.0456,
      "step": 13406
    },
    {
      "epoch": 0.9664095725509984,
      "grad_norm": 0.14430981351380665,
      "learning_rate": 5.912636643853109e-07,
      "loss": 1.0127,
      "step": 13407
    },
    {
      "epoch": 0.9664816550133353,
      "grad_norm": 0.11936890265925759,
      "learning_rate": 5.887312650729281e-07,
      "loss": 1.0929,
      "step": 13408
    },
    {
      "epoch": 0.9665537374756722,
      "grad_norm": 0.12095580320652038,
      "learning_rate": 5.862042845640403e-07,
      "loss": 1.0309,
      "step": 13409
    },
    {
      "epoch": 0.9666258199380091,
      "grad_norm": 0.17534671432229412,
      "learning_rate": 5.836827229964037e-07,
      "loss": 1.064,
      "step": 13410
    },
    {
      "epoch": 0.966697902400346,
      "grad_norm": 0.13025636721011435,
      "learning_rate": 5.811665805074639e-07,
      "loss": 1.0463,
      "step": 13411
    },
    {
      "epoch": 0.966769984862683,
      "grad_norm": 0.32075785325812467,
      "learning_rate": 5.78655857234367e-07,
      "loss": 1.0955,
      "step": 13412
    },
    {
      "epoch": 0.9668420673250199,
      "grad_norm": 0.16063178655353008,
      "learning_rate": 5.76150553313981e-07,
      "loss": 1.0633,
      "step": 13413
    },
    {
      "epoch": 0.9669141497873568,
      "grad_norm": 0.10839864946235249,
      "learning_rate": 5.736506688828636e-07,
      "loss": 0.9983,
      "step": 13414
    },
    {
      "epoch": 0.9669862322496936,
      "grad_norm": 0.12565271436018213,
      "learning_rate": 5.711562040772723e-07,
      "loss": 1.0474,
      "step": 13415
    },
    {
      "epoch": 0.9670583147120305,
      "grad_norm": 0.13160587894147177,
      "learning_rate": 5.686671590331871e-07,
      "loss": 0.9946,
      "step": 13416
    },
    {
      "epoch": 0.9671303971743674,
      "grad_norm": 0.13829448651974313,
      "learning_rate": 5.661835338862775e-07,
      "loss": 1.0332,
      "step": 13417
    },
    {
      "epoch": 0.9672024796367044,
      "grad_norm": 0.14405354393808736,
      "learning_rate": 5.63705328771924e-07,
      "loss": 1.1026,
      "step": 13418
    },
    {
      "epoch": 0.9672745620990413,
      "grad_norm": 0.1345673931806746,
      "learning_rate": 5.612325438252186e-07,
      "loss": 1.0696,
      "step": 13419
    },
    {
      "epoch": 0.9673466445613782,
      "grad_norm": 0.13172092787279607,
      "learning_rate": 5.587651791809312e-07,
      "loss": 1.0485,
      "step": 13420
    },
    {
      "epoch": 0.9674187270237151,
      "grad_norm": 0.24712982208289658,
      "learning_rate": 5.563032349735875e-07,
      "loss": 0.9671,
      "step": 13421
    },
    {
      "epoch": 0.967490809486052,
      "grad_norm": 0.12733017194172314,
      "learning_rate": 5.538467113373469e-07,
      "loss": 1.079,
      "step": 13422
    },
    {
      "epoch": 0.967562891948389,
      "grad_norm": 0.116256217681441,
      "learning_rate": 5.513956084061467e-07,
      "loss": 1.0573,
      "step": 13423
    },
    {
      "epoch": 0.9676349744107259,
      "grad_norm": 0.13912006659792378,
      "learning_rate": 5.48949926313569e-07,
      "loss": 1.0444,
      "step": 13424
    },
    {
      "epoch": 0.9677070568730628,
      "grad_norm": 0.10641355777989951,
      "learning_rate": 5.465096651929402e-07,
      "loss": 0.9422,
      "step": 13425
    },
    {
      "epoch": 0.9677791393353997,
      "grad_norm": 0.12680375081449533,
      "learning_rate": 5.440748251772764e-07,
      "loss": 1.0558,
      "step": 13426
    },
    {
      "epoch": 0.9678512217977366,
      "grad_norm": 0.13780886345746654,
      "learning_rate": 5.416454063992715e-07,
      "loss": 1.0618,
      "step": 13427
    },
    {
      "epoch": 0.9679233042600736,
      "grad_norm": 0.2126661725497374,
      "learning_rate": 5.392214089913861e-07,
      "loss": 1.0063,
      "step": 13428
    },
    {
      "epoch": 0.9679953867224105,
      "grad_norm": 0.13606278903804886,
      "learning_rate": 5.368028330857367e-07,
      "loss": 1.0456,
      "step": 13429
    },
    {
      "epoch": 0.9680674691847474,
      "grad_norm": 0.10961141191220326,
      "learning_rate": 5.343896788141622e-07,
      "loss": 1.0362,
      "step": 13430
    },
    {
      "epoch": 0.9681395516470843,
      "grad_norm": 0.13594935268309302,
      "learning_rate": 5.319819463081799e-07,
      "loss": 1.0728,
      "step": 13431
    },
    {
      "epoch": 0.9682116341094211,
      "grad_norm": 0.1287857844420859,
      "learning_rate": 5.295796356990512e-07,
      "loss": 1.0198,
      "step": 13432
    },
    {
      "epoch": 0.968283716571758,
      "grad_norm": 0.13477676443235512,
      "learning_rate": 5.271827471177271e-07,
      "loss": 0.9609,
      "step": 13433
    },
    {
      "epoch": 0.968355799034095,
      "grad_norm": 0.12149016037601458,
      "learning_rate": 5.247912806948585e-07,
      "loss": 1.0248,
      "step": 13434
    },
    {
      "epoch": 0.9684278814964319,
      "grad_norm": 0.12367174759640451,
      "learning_rate": 5.224052365607746e-07,
      "loss": 0.9694,
      "step": 13435
    },
    {
      "epoch": 0.9684999639587688,
      "grad_norm": 0.1550302392459402,
      "learning_rate": 5.200246148455712e-07,
      "loss": 1.0769,
      "step": 13436
    },
    {
      "epoch": 0.9685720464211057,
      "grad_norm": 0.1276708559843585,
      "learning_rate": 5.176494156790002e-07,
      "loss": 1.0271,
      "step": 13437
    },
    {
      "epoch": 0.9686441288834426,
      "grad_norm": 0.1342934186587719,
      "learning_rate": 5.152796391905357e-07,
      "loss": 1.0822,
      "step": 13438
    },
    {
      "epoch": 0.9687162113457796,
      "grad_norm": 0.13062900666467442,
      "learning_rate": 5.1291528550933e-07,
      "loss": 1.0288,
      "step": 13439
    },
    {
      "epoch": 0.9687882938081165,
      "grad_norm": 0.1369899358310802,
      "learning_rate": 5.105563547642911e-07,
      "loss": 1.1559,
      "step": 13440
    },
    {
      "epoch": 0.9688603762704534,
      "grad_norm": 0.18025628305099545,
      "learning_rate": 5.082028470839828e-07,
      "loss": 1.0634,
      "step": 13441
    },
    {
      "epoch": 0.9689324587327903,
      "grad_norm": 0.12233977664765922,
      "learning_rate": 5.058547625967025e-07,
      "loss": 1.0664,
      "step": 13442
    },
    {
      "epoch": 0.9690045411951272,
      "grad_norm": 0.12106567765237028,
      "learning_rate": 5.035121014304367e-07,
      "loss": 0.9578,
      "step": 13443
    },
    {
      "epoch": 0.9690766236574642,
      "grad_norm": 0.13057118752649788,
      "learning_rate": 5.011748637128722e-07,
      "loss": 1.0258,
      "step": 13444
    },
    {
      "epoch": 0.9691487061198011,
      "grad_norm": 0.11491586967165417,
      "learning_rate": 4.988430495714292e-07,
      "loss": 0.9524,
      "step": 13445
    },
    {
      "epoch": 0.969220788582138,
      "grad_norm": 0.11958725520349389,
      "learning_rate": 4.96516659133206e-07,
      "loss": 0.961,
      "step": 13446
    },
    {
      "epoch": 0.9692928710444749,
      "grad_norm": 0.12846472725762736,
      "learning_rate": 4.941956925250013e-07,
      "loss": 0.9959,
      "step": 13447
    },
    {
      "epoch": 0.9693649535068118,
      "grad_norm": 0.13302973380588154,
      "learning_rate": 4.918801498733472e-07,
      "loss": 0.9891,
      "step": 13448
    },
    {
      "epoch": 0.9694370359691487,
      "grad_norm": 0.1201400386193862,
      "learning_rate": 4.895700313044315e-07,
      "loss": 1.037,
      "step": 13449
    },
    {
      "epoch": 0.9695091184314856,
      "grad_norm": 0.16645819482882307,
      "learning_rate": 4.872653369442092e-07,
      "loss": 1.0053,
      "step": 13450
    },
    {
      "epoch": 0.9695812008938225,
      "grad_norm": 0.12136458763224423,
      "learning_rate": 4.849660669182909e-07,
      "loss": 1.0541,
      "step": 13451
    },
    {
      "epoch": 0.9696532833561594,
      "grad_norm": 0.1450073278056433,
      "learning_rate": 4.826722213520096e-07,
      "loss": 1.0903,
      "step": 13452
    },
    {
      "epoch": 0.9697253658184963,
      "grad_norm": 0.11450529690556321,
      "learning_rate": 4.803838003703876e-07,
      "loss": 1.0523,
      "step": 13453
    },
    {
      "epoch": 0.9697974482808333,
      "grad_norm": 0.13084474943177954,
      "learning_rate": 4.781008040981915e-07,
      "loss": 0.9839,
      "step": 13454
    },
    {
      "epoch": 0.9698695307431702,
      "grad_norm": 0.14592049465344797,
      "learning_rate": 4.758232326598444e-07,
      "loss": 1.0122,
      "step": 13455
    },
    {
      "epoch": 0.9699416132055071,
      "grad_norm": 0.17661391166980425,
      "learning_rate": 4.735510861795023e-07,
      "loss": 1.0097,
      "step": 13456
    },
    {
      "epoch": 0.970013695667844,
      "grad_norm": 0.13885333542186748,
      "learning_rate": 4.7128436478099945e-07,
      "loss": 1.0616,
      "step": 13457
    },
    {
      "epoch": 0.9700857781301809,
      "grad_norm": 0.13237985314626072,
      "learning_rate": 4.69023068587926e-07,
      "loss": 1.0147,
      "step": 13458
    },
    {
      "epoch": 0.9701578605925179,
      "grad_norm": 0.20564570827102202,
      "learning_rate": 4.6676719772352775e-07,
      "loss": 1.0923,
      "step": 13459
    },
    {
      "epoch": 0.9702299430548548,
      "grad_norm": 0.12394077821886408,
      "learning_rate": 4.6451675231075077e-07,
      "loss": 1.0969,
      "step": 13460
    },
    {
      "epoch": 0.9703020255171917,
      "grad_norm": 0.11185351508162256,
      "learning_rate": 4.6227173247228583e-07,
      "loss": 1.0555,
      "step": 13461
    },
    {
      "epoch": 0.9703741079795286,
      "grad_norm": 0.12066497596289401,
      "learning_rate": 4.6003213833051284e-07,
      "loss": 1.0253,
      "step": 13462
    },
    {
      "epoch": 0.9704461904418655,
      "grad_norm": 0.11513168947959511,
      "learning_rate": 4.577979700074897e-07,
      "loss": 1.008,
      "step": 13463
    },
    {
      "epoch": 0.9705182729042025,
      "grad_norm": 0.11730083761244217,
      "learning_rate": 4.555692276250079e-07,
      "loss": 1.0618,
      "step": 13464
    },
    {
      "epoch": 0.9705903553665394,
      "grad_norm": 0.12377848990019753,
      "learning_rate": 4.5334591130454795e-07,
      "loss": 1.0477,
      "step": 13465
    },
    {
      "epoch": 0.9706624378288762,
      "grad_norm": 0.11216364863256403,
      "learning_rate": 4.51128021167313e-07,
      "loss": 0.9858,
      "step": 13466
    },
    {
      "epoch": 0.9707345202912131,
      "grad_norm": 0.12967185447375645,
      "learning_rate": 4.489155573341841e-07,
      "loss": 1.0081,
      "step": 13467
    },
    {
      "epoch": 0.97080660275355,
      "grad_norm": 0.18069011775394983,
      "learning_rate": 4.467085199257759e-07,
      "loss": 1.0425,
      "step": 13468
    },
    {
      "epoch": 0.9708786852158869,
      "grad_norm": 0.12644923085080712,
      "learning_rate": 4.445069090623699e-07,
      "loss": 1.0356,
      "step": 13469
    },
    {
      "epoch": 0.9709507676782239,
      "grad_norm": 0.12484160005968485,
      "learning_rate": 4.423107248639924e-07,
      "loss": 1.0139,
      "step": 13470
    },
    {
      "epoch": 0.9710228501405608,
      "grad_norm": 0.11342395445707784,
      "learning_rate": 4.4011996745034755e-07,
      "loss": 0.9428,
      "step": 13471
    },
    {
      "epoch": 0.9710949326028977,
      "grad_norm": 0.13306047134690127,
      "learning_rate": 4.379346369408621e-07,
      "loss": 0.9647,
      "step": 13472
    },
    {
      "epoch": 0.9711670150652346,
      "grad_norm": 0.13383434895067853,
      "learning_rate": 4.357547334546408e-07,
      "loss": 1.0377,
      "step": 13473
    },
    {
      "epoch": 0.9712390975275715,
      "grad_norm": 0.13245330925377227,
      "learning_rate": 4.335802571105108e-07,
      "loss": 1.0286,
      "step": 13474
    },
    {
      "epoch": 0.9713111799899085,
      "grad_norm": 0.11995185145146608,
      "learning_rate": 4.314112080269994e-07,
      "loss": 1.0327,
      "step": 13475
    },
    {
      "epoch": 0.9713832624522454,
      "grad_norm": 0.1167269567357976,
      "learning_rate": 4.292475863223455e-07,
      "loss": 0.9337,
      "step": 13476
    },
    {
      "epoch": 0.9714553449145823,
      "grad_norm": 0.10538940770993335,
      "learning_rate": 4.2708939211448806e-07,
      "loss": 1.1395,
      "step": 13477
    },
    {
      "epoch": 0.9715274273769192,
      "grad_norm": 0.1291244020651031,
      "learning_rate": 4.2493662552106627e-07,
      "loss": 1.0598,
      "step": 13478
    },
    {
      "epoch": 0.9715995098392561,
      "grad_norm": 0.11504003953905019,
      "learning_rate": 4.2278928665941964e-07,
      "loss": 1.0609,
      "step": 13479
    },
    {
      "epoch": 0.9716715923015931,
      "grad_norm": 0.15466179558214757,
      "learning_rate": 4.2064737564659896e-07,
      "loss": 1.0347,
      "step": 13480
    },
    {
      "epoch": 0.97174367476393,
      "grad_norm": 0.11508198363360346,
      "learning_rate": 4.185108925993553e-07,
      "loss": 1.0784,
      "step": 13481
    },
    {
      "epoch": 0.9718157572262669,
      "grad_norm": 0.3159483564220924,
      "learning_rate": 4.16379837634151e-07,
      "loss": 1.0962,
      "step": 13482
    },
    {
      "epoch": 0.9718878396886038,
      "grad_norm": 0.121666920290873,
      "learning_rate": 4.142542108671599e-07,
      "loss": 1.012,
      "step": 13483
    },
    {
      "epoch": 0.9719599221509406,
      "grad_norm": 0.14362519778027918,
      "learning_rate": 4.1213401241421147e-07,
      "loss": 1.0571,
      "step": 13484
    },
    {
      "epoch": 0.9720320046132775,
      "grad_norm": 0.13005596932925867,
      "learning_rate": 4.100192423909133e-07,
      "loss": 1.0002,
      "step": 13485
    },
    {
      "epoch": 0.9721040870756145,
      "grad_norm": 0.13175739796637154,
      "learning_rate": 4.0790990091251755e-07,
      "loss": 0.9815,
      "step": 13486
    },
    {
      "epoch": 0.9721761695379514,
      "grad_norm": 0.11344542427252961,
      "learning_rate": 4.0580598809401013e-07,
      "loss": 1.0604,
      "step": 13487
    },
    {
      "epoch": 0.9722482520002883,
      "grad_norm": 0.12973539581018817,
      "learning_rate": 4.0370750405006595e-07,
      "loss": 1.0235,
      "step": 13488
    },
    {
      "epoch": 0.9723203344626252,
      "grad_norm": 0.12341311727370854,
      "learning_rate": 4.0161444889508237e-07,
      "loss": 1.0616,
      "step": 13489
    },
    {
      "epoch": 0.9723924169249621,
      "grad_norm": 0.2514369900825998,
      "learning_rate": 3.995268227431459e-07,
      "loss": 1.0537,
      "step": 13490
    },
    {
      "epoch": 0.9724644993872991,
      "grad_norm": 0.12847909993646772,
      "learning_rate": 3.974446257080433e-07,
      "loss": 1.091,
      "step": 13491
    },
    {
      "epoch": 0.972536581849636,
      "grad_norm": 0.12291743593344527,
      "learning_rate": 3.953678579032838e-07,
      "loss": 1.0139,
      "step": 13492
    },
    {
      "epoch": 0.9726086643119729,
      "grad_norm": 0.1588548552717254,
      "learning_rate": 3.9329651944205456e-07,
      "loss": 1.0319,
      "step": 13493
    },
    {
      "epoch": 0.9726807467743098,
      "grad_norm": 0.14117319070488546,
      "learning_rate": 3.9123061043728757e-07,
      "loss": 0.9898,
      "step": 13494
    },
    {
      "epoch": 0.9727528292366467,
      "grad_norm": 0.12981372908393976,
      "learning_rate": 3.8917013100157054e-07,
      "loss": 1.029,
      "step": 13495
    },
    {
      "epoch": 0.9728249116989837,
      "grad_norm": 0.1496745798177582,
      "learning_rate": 3.871150812472246e-07,
      "loss": 1.03,
      "step": 13496
    },
    {
      "epoch": 0.9728969941613206,
      "grad_norm": 0.1342648585035089,
      "learning_rate": 3.8506546128626034e-07,
      "loss": 0.9675,
      "step": 13497
    },
    {
      "epoch": 0.9729690766236575,
      "grad_norm": 0.11521061132550248,
      "learning_rate": 3.8302127123042155e-07,
      "loss": 1.0269,
      "step": 13498
    },
    {
      "epoch": 0.9730411590859944,
      "grad_norm": 0.13702919929880647,
      "learning_rate": 3.809825111911192e-07,
      "loss": 1.0616,
      "step": 13499
    },
    {
      "epoch": 0.9731132415483313,
      "grad_norm": 0.1899633907104594,
      "learning_rate": 3.7894918127948655e-07,
      "loss": 0.9487,
      "step": 13500
    },
    {
      "epoch": 0.9731853240106682,
      "grad_norm": 0.11360747131394675,
      "learning_rate": 3.769212816063572e-07,
      "loss": 1.0428,
      "step": 13501
    },
    {
      "epoch": 0.9732574064730051,
      "grad_norm": 0.11768486327612462,
      "learning_rate": 3.748988122822761e-07,
      "loss": 1.0277,
      "step": 13502
    },
    {
      "epoch": 0.973329488935342,
      "grad_norm": 0.10862607950346671,
      "learning_rate": 3.728817734174772e-07,
      "loss": 1.0332,
      "step": 13503
    },
    {
      "epoch": 0.9734015713976789,
      "grad_norm": 0.13487100774987618,
      "learning_rate": 3.7087016512191705e-07,
      "loss": 1.0408,
      "step": 13504
    },
    {
      "epoch": 0.9734736538600158,
      "grad_norm": 0.13860462239391863,
      "learning_rate": 3.688639875052302e-07,
      "loss": 1.0394,
      "step": 13505
    },
    {
      "epoch": 0.9735457363223528,
      "grad_norm": 0.11314458487779681,
      "learning_rate": 3.668632406767958e-07,
      "loss": 1.0834,
      "step": 13506
    },
    {
      "epoch": 0.9736178187846897,
      "grad_norm": 0.15213383355283575,
      "learning_rate": 3.648679247456488e-07,
      "loss": 1.0108,
      "step": 13507
    },
    {
      "epoch": 0.9736899012470266,
      "grad_norm": 0.13444133836722214,
      "learning_rate": 3.628780398205578e-07,
      "loss": 1.052,
      "step": 13508
    },
    {
      "epoch": 0.9737619837093635,
      "grad_norm": 0.14684779293772288,
      "learning_rate": 3.6089358601000265e-07,
      "loss": 0.9331,
      "step": 13509
    },
    {
      "epoch": 0.9738340661717004,
      "grad_norm": 0.14850423958905312,
      "learning_rate": 3.589145634221414e-07,
      "loss": 1.0304,
      "step": 13510
    },
    {
      "epoch": 0.9739061486340373,
      "grad_norm": 0.12335341992891721,
      "learning_rate": 3.5694097216484314e-07,
      "loss": 0.976,
      "step": 13511
    },
    {
      "epoch": 0.9739782310963743,
      "grad_norm": 0.12604666910881068,
      "learning_rate": 3.5497281234569965e-07,
      "loss": 1.0686,
      "step": 13512
    },
    {
      "epoch": 0.9740503135587112,
      "grad_norm": 0.11021324838291782,
      "learning_rate": 3.5301008407199187e-07,
      "loss": 1.1299,
      "step": 13513
    },
    {
      "epoch": 0.9741223960210481,
      "grad_norm": 0.1183003721190956,
      "learning_rate": 3.510527874506897e-07,
      "loss": 1.0697,
      "step": 13514
    },
    {
      "epoch": 0.974194478483385,
      "grad_norm": 0.13686952994468846,
      "learning_rate": 3.491009225885078e-07,
      "loss": 1.0401,
      "step": 13515
    },
    {
      "epoch": 0.974266560945722,
      "grad_norm": 0.12417182356107438,
      "learning_rate": 3.471544895918277e-07,
      "loss": 1.059,
      "step": 13516
    },
    {
      "epoch": 0.9743386434080589,
      "grad_norm": 0.12863058817374304,
      "learning_rate": 3.452134885667424e-07,
      "loss": 1.0642,
      "step": 13517
    },
    {
      "epoch": 0.9744107258703957,
      "grad_norm": 0.1188344112148927,
      "learning_rate": 3.432779196190561e-07,
      "loss": 1.0898,
      "step": 13518
    },
    {
      "epoch": 0.9744828083327326,
      "grad_norm": 0.11822634571952728,
      "learning_rate": 3.4134778285428437e-07,
      "loss": 1.0223,
      "step": 13519
    },
    {
      "epoch": 0.9745548907950695,
      "grad_norm": 1.2964161127621106,
      "learning_rate": 3.39423078377632e-07,
      "loss": 1.076,
      "step": 13520
    },
    {
      "epoch": 0.9746269732574064,
      "grad_norm": 0.10844058668147773,
      "learning_rate": 3.375038062940039e-07,
      "loss": 0.9718,
      "step": 13521
    },
    {
      "epoch": 0.9746990557197434,
      "grad_norm": 0.16195020724048143,
      "learning_rate": 3.3558996670802757e-07,
      "loss": 1.0699,
      "step": 13522
    },
    {
      "epoch": 0.9747711381820803,
      "grad_norm": 0.13024553111562576,
      "learning_rate": 3.3368155972401946e-07,
      "loss": 1.0515,
      "step": 13523
    },
    {
      "epoch": 0.9748432206444172,
      "grad_norm": 0.12802632496450284,
      "learning_rate": 3.317785854460076e-07,
      "loss": 1.0545,
      "step": 13524
    },
    {
      "epoch": 0.9749153031067541,
      "grad_norm": 0.13749396213614973,
      "learning_rate": 3.298810439777311e-07,
      "loss": 0.9896,
      "step": 13525
    },
    {
      "epoch": 0.974987385569091,
      "grad_norm": 0.202312047766685,
      "learning_rate": 3.2798893542259624e-07,
      "loss": 0.9568,
      "step": 13526
    },
    {
      "epoch": 0.975059468031428,
      "grad_norm": 0.12274198281159476,
      "learning_rate": 3.261022598837649e-07,
      "loss": 1.0537,
      "step": 13527
    },
    {
      "epoch": 0.9751315504937649,
      "grad_norm": 0.11101057113899214,
      "learning_rate": 3.2422101746406586e-07,
      "loss": 0.9727,
      "step": 13528
    },
    {
      "epoch": 0.9752036329561018,
      "grad_norm": 0.1976205979626967,
      "learning_rate": 3.223452082660394e-07,
      "loss": 1.1151,
      "step": 13529
    },
    {
      "epoch": 0.9752757154184387,
      "grad_norm": 0.14258996918766148,
      "learning_rate": 3.204748323919482e-07,
      "loss": 1.0239,
      "step": 13530
    },
    {
      "epoch": 0.9753477978807756,
      "grad_norm": 0.1912573429243692,
      "learning_rate": 3.18609889943744e-07,
      "loss": 1.0078,
      "step": 13531
    },
    {
      "epoch": 0.9754198803431126,
      "grad_norm": 0.12206587032873141,
      "learning_rate": 3.1675038102306765e-07,
      "loss": 0.9949,
      "step": 13532
    },
    {
      "epoch": 0.9754919628054495,
      "grad_norm": 0.12551778483647522,
      "learning_rate": 3.148963057312826e-07,
      "loss": 1.0365,
      "step": 13533
    },
    {
      "epoch": 0.9755640452677864,
      "grad_norm": 0.12082804891356753,
      "learning_rate": 3.130476641694635e-07,
      "loss": 1.0558,
      "step": 13534
    },
    {
      "epoch": 0.9756361277301233,
      "grad_norm": 0.12635749329893692,
      "learning_rate": 3.112044564383743e-07,
      "loss": 0.9651,
      "step": 13535
    },
    {
      "epoch": 0.9757082101924601,
      "grad_norm": 0.20079920180248764,
      "learning_rate": 3.093666826384789e-07,
      "loss": 1.027,
      "step": 13536
    },
    {
      "epoch": 0.975780292654797,
      "grad_norm": 0.12375863488926939,
      "learning_rate": 3.0753434286995284e-07,
      "loss": 0.9777,
      "step": 13537
    },
    {
      "epoch": 0.975852375117134,
      "grad_norm": 0.17591841178755038,
      "learning_rate": 3.05707437232694e-07,
      "loss": 1.0666,
      "step": 13538
    },
    {
      "epoch": 0.9759244575794709,
      "grad_norm": 0.12466078922474823,
      "learning_rate": 3.0388596582625605e-07,
      "loss": 1.0414,
      "step": 13539
    },
    {
      "epoch": 0.9759965400418078,
      "grad_norm": 1.3550417074067447,
      "learning_rate": 3.020699287499484e-07,
      "loss": 0.9793,
      "step": 13540
    },
    {
      "epoch": 0.9760686225041447,
      "grad_norm": 0.11065789990208741,
      "learning_rate": 3.0025932610274754e-07,
      "loss": 1.0428,
      "step": 13541
    },
    {
      "epoch": 0.9761407049664816,
      "grad_norm": 0.157430451651222,
      "learning_rate": 2.984541579833522e-07,
      "loss": 1.0407,
      "step": 13542
    },
    {
      "epoch": 0.9762127874288186,
      "grad_norm": 0.12114822927223108,
      "learning_rate": 2.9665442449017257e-07,
      "loss": 1.0309,
      "step": 13543
    },
    {
      "epoch": 0.9762848698911555,
      "grad_norm": 0.12784250186298635,
      "learning_rate": 2.9486012572128573e-07,
      "loss": 1.0207,
      "step": 13544
    },
    {
      "epoch": 0.9763569523534924,
      "grad_norm": 0.17892584110011914,
      "learning_rate": 2.930712617745135e-07,
      "loss": 1.0244,
      "step": 13545
    },
    {
      "epoch": 0.9764290348158293,
      "grad_norm": 0.11602954497886643,
      "learning_rate": 2.912878327473556e-07,
      "loss": 0.9665,
      "step": 13546
    },
    {
      "epoch": 0.9765011172781662,
      "grad_norm": 0.1269986168104009,
      "learning_rate": 2.895098387370454e-07,
      "loss": 0.9685,
      "step": 13547
    },
    {
      "epoch": 0.9765731997405032,
      "grad_norm": 0.11471160218498654,
      "learning_rate": 2.87737279840472e-07,
      "loss": 0.9924,
      "step": 13548
    },
    {
      "epoch": 0.9766452822028401,
      "grad_norm": 0.13023060678690918,
      "learning_rate": 2.8597015615426937e-07,
      "loss": 1.0612,
      "step": 13549
    },
    {
      "epoch": 0.976717364665177,
      "grad_norm": 0.15162295031756873,
      "learning_rate": 2.842084677747603e-07,
      "loss": 0.9968,
      "step": 13550
    },
    {
      "epoch": 0.9767894471275139,
      "grad_norm": 0.14326410689500801,
      "learning_rate": 2.8245221479797913e-07,
      "loss": 1.0679,
      "step": 13551
    },
    {
      "epoch": 0.9768615295898508,
      "grad_norm": 0.13317631203126615,
      "learning_rate": 2.807013973196493e-07,
      "loss": 1.0652,
      "step": 13552
    },
    {
      "epoch": 0.9769336120521876,
      "grad_norm": 0.11954195061491173,
      "learning_rate": 2.7895601543520553e-07,
      "loss": 1.0391,
      "step": 13553
    },
    {
      "epoch": 0.9770056945145246,
      "grad_norm": 0.129239084247838,
      "learning_rate": 2.7721606923978293e-07,
      "loss": 1.0422,
      "step": 13554
    },
    {
      "epoch": 0.9770777769768615,
      "grad_norm": 0.1352456668203093,
      "learning_rate": 2.754815588282389e-07,
      "loss": 1.0937,
      "step": 13555
    },
    {
      "epoch": 0.9771498594391984,
      "grad_norm": 0.11239876214600439,
      "learning_rate": 2.737524842951089e-07,
      "loss": 1.0661,
      "step": 13556
    },
    {
      "epoch": 0.9772219419015353,
      "grad_norm": 0.13711939188581604,
      "learning_rate": 2.720288457346398e-07,
      "loss": 1.0045,
      "step": 13557
    },
    {
      "epoch": 0.9772940243638722,
      "grad_norm": 0.12249048777662266,
      "learning_rate": 2.703106432408009e-07,
      "loss": 0.9956,
      "step": 13558
    },
    {
      "epoch": 0.9773661068262092,
      "grad_norm": 0.11611530275713494,
      "learning_rate": 2.685978769072284e-07,
      "loss": 1.001,
      "step": 13559
    },
    {
      "epoch": 0.9774381892885461,
      "grad_norm": 0.13486794762372306,
      "learning_rate": 2.6689054682730306e-07,
      "loss": 0.9989,
      "step": 13560
    },
    {
      "epoch": 0.977510271750883,
      "grad_norm": 0.15991140104522353,
      "learning_rate": 2.651886530940728e-07,
      "loss": 0.9726,
      "step": 13561
    },
    {
      "epoch": 0.9775823542132199,
      "grad_norm": 0.12940199262581925,
      "learning_rate": 2.6349219580031893e-07,
      "loss": 1.0359,
      "step": 13562
    },
    {
      "epoch": 0.9776544366755568,
      "grad_norm": 0.1525742019091484,
      "learning_rate": 2.6180117503850077e-07,
      "loss": 1.0085,
      "step": 13563
    },
    {
      "epoch": 0.9777265191378938,
      "grad_norm": 1.0589850963910372,
      "learning_rate": 2.601155909008113e-07,
      "loss": 1.0959,
      "step": 13564
    },
    {
      "epoch": 0.9777986016002307,
      "grad_norm": 0.11819778469810067,
      "learning_rate": 2.584354434791214e-07,
      "loss": 0.9515,
      "step": 13565
    },
    {
      "epoch": 0.9778706840625676,
      "grad_norm": 0.12023545787031899,
      "learning_rate": 2.5676073286500237e-07,
      "loss": 0.9857,
      "step": 13566
    },
    {
      "epoch": 0.9779427665249045,
      "grad_norm": 0.11300599041556328,
      "learning_rate": 2.5509145914975885e-07,
      "loss": 0.9937,
      "step": 13567
    },
    {
      "epoch": 0.9780148489872414,
      "grad_norm": 0.23487145677960028,
      "learning_rate": 2.534276224243848e-07,
      "loss": 1.0351,
      "step": 13568
    },
    {
      "epoch": 0.9780869314495784,
      "grad_norm": 0.12854611912211272,
      "learning_rate": 2.517692227795521e-07,
      "loss": 1.0191,
      "step": 13569
    },
    {
      "epoch": 0.9781590139119152,
      "grad_norm": 0.12268296134610021,
      "learning_rate": 2.5011626030566614e-07,
      "loss": 1.0436,
      "step": 13570
    },
    {
      "epoch": 0.9782310963742521,
      "grad_norm": 0.11424173470945011,
      "learning_rate": 2.4846873509284383e-07,
      "loss": 1.0405,
      "step": 13571
    },
    {
      "epoch": 0.978303178836589,
      "grad_norm": 0.13185238444398298,
      "learning_rate": 2.468266472308689e-07,
      "loss": 0.9449,
      "step": 13572
    },
    {
      "epoch": 0.9783752612989259,
      "grad_norm": 0.13917435057179908,
      "learning_rate": 2.451899968092586e-07,
      "loss": 0.9852,
      "step": 13573
    },
    {
      "epoch": 0.9784473437612629,
      "grad_norm": 0.11972131381269906,
      "learning_rate": 2.435587839172304e-07,
      "loss": 1.0688,
      "step": 13574
    },
    {
      "epoch": 0.9785194262235998,
      "grad_norm": 0.13353326559520792,
      "learning_rate": 2.4193300864369104e-07,
      "loss": 1.03,
      "step": 13575
    },
    {
      "epoch": 0.9785915086859367,
      "grad_norm": 0.136713560336243,
      "learning_rate": 2.4031267107726964e-07,
      "loss": 1.1023,
      "step": 13576
    },
    {
      "epoch": 0.9786635911482736,
      "grad_norm": 0.10409000351841198,
      "learning_rate": 2.3869777130627324e-07,
      "loss": 1.0506,
      "step": 13577
    },
    {
      "epoch": 0.9787356736106105,
      "grad_norm": 0.10675204877791732,
      "learning_rate": 2.370883094187537e-07,
      "loss": 1.0481,
      "step": 13578
    },
    {
      "epoch": 0.9788077560729475,
      "grad_norm": 0.12066094340533542,
      "learning_rate": 2.3548428550241865e-07,
      "loss": 1.0614,
      "step": 13579
    },
    {
      "epoch": 0.9788798385352844,
      "grad_norm": 0.12575788286446382,
      "learning_rate": 2.3388569964470918e-07,
      "loss": 0.9583,
      "step": 13580
    },
    {
      "epoch": 0.9789519209976213,
      "grad_norm": 0.11062991956622532,
      "learning_rate": 2.3229255193275569e-07,
      "loss": 1.0515,
      "step": 13581
    },
    {
      "epoch": 0.9790240034599582,
      "grad_norm": 0.12369489802467648,
      "learning_rate": 2.3070484245341084e-07,
      "loss": 1.086,
      "step": 13582
    },
    {
      "epoch": 0.9790960859222951,
      "grad_norm": 0.12416447082334592,
      "learning_rate": 2.2912257129320547e-07,
      "loss": 1.0611,
      "step": 13583
    },
    {
      "epoch": 0.979168168384632,
      "grad_norm": 0.15236004316818821,
      "learning_rate": 2.2754573853841498e-07,
      "loss": 1.0068,
      "step": 13584
    },
    {
      "epoch": 0.979240250846969,
      "grad_norm": 0.12046796951589242,
      "learning_rate": 2.2597434427495956e-07,
      "loss": 1.0435,
      "step": 13585
    },
    {
      "epoch": 0.9793123333093059,
      "grad_norm": 0.12249714184279185,
      "learning_rate": 2.24408388588504e-07,
      "loss": 0.9956,
      "step": 13586
    },
    {
      "epoch": 0.9793844157716428,
      "grad_norm": 0.12210304161225838,
      "learning_rate": 2.2284787156441333e-07,
      "loss": 1.0183,
      "step": 13587
    },
    {
      "epoch": 0.9794564982339796,
      "grad_norm": 0.11783987014065705,
      "learning_rate": 2.2129279328774176e-07,
      "loss": 0.9878,
      "step": 13588
    },
    {
      "epoch": 0.9795285806963165,
      "grad_norm": 0.12296586485295266,
      "learning_rate": 2.1974315384325485e-07,
      "loss": 1.0472,
      "step": 13589
    },
    {
      "epoch": 0.9796006631586535,
      "grad_norm": 0.12482996833445724,
      "learning_rate": 2.181989533154405e-07,
      "loss": 1.0265,
      "step": 13590
    },
    {
      "epoch": 0.9796727456209904,
      "grad_norm": 0.12375803479215731,
      "learning_rate": 2.1666019178844255e-07,
      "loss": 1.1007,
      "step": 13591
    },
    {
      "epoch": 0.9797448280833273,
      "grad_norm": 0.12352405998834669,
      "learning_rate": 2.1512686934616055e-07,
      "loss": 0.9921,
      "step": 13592
    },
    {
      "epoch": 0.9798169105456642,
      "grad_norm": 0.13527255179454556,
      "learning_rate": 2.13598986072161e-07,
      "loss": 0.9886,
      "step": 13593
    },
    {
      "epoch": 0.9798889930080011,
      "grad_norm": 0.15060889878493747,
      "learning_rate": 2.120765420497328e-07,
      "loss": 1.095,
      "step": 13594
    },
    {
      "epoch": 0.9799610754703381,
      "grad_norm": 0.11738583252367764,
      "learning_rate": 2.1055953736186517e-07,
      "loss": 1.0562,
      "step": 13595
    },
    {
      "epoch": 0.980033157932675,
      "grad_norm": 0.15079349786229646,
      "learning_rate": 2.090479720912364e-07,
      "loss": 0.9966,
      "step": 13596
    },
    {
      "epoch": 0.9801052403950119,
      "grad_norm": 0.11283504838009274,
      "learning_rate": 2.075418463202583e-07,
      "loss": 1.0623,
      "step": 13597
    },
    {
      "epoch": 0.9801773228573488,
      "grad_norm": 0.12522223631721785,
      "learning_rate": 2.0604116013100972e-07,
      "loss": 0.9579,
      "step": 13598
    },
    {
      "epoch": 0.9802494053196857,
      "grad_norm": 0.1324115880375863,
      "learning_rate": 2.0454591360530297e-07,
      "loss": 1.0138,
      "step": 13599
    },
    {
      "epoch": 0.9803214877820227,
      "grad_norm": 0.121435158884467,
      "learning_rate": 2.030561068246395e-07,
      "loss": 1.0655,
      "step": 13600
    },
    {
      "epoch": 0.9803935702443596,
      "grad_norm": 0.13615731393026556,
      "learning_rate": 2.0157173987022104e-07,
      "loss": 0.9742,
      "step": 13601
    },
    {
      "epoch": 0.9804656527066965,
      "grad_norm": 0.12019065997874491,
      "learning_rate": 2.0009281282297176e-07,
      "loss": 1.0215,
      "step": 13602
    },
    {
      "epoch": 0.9805377351690334,
      "grad_norm": 0.12165785180804334,
      "learning_rate": 1.9861932576350494e-07,
      "loss": 1.0367,
      "step": 13603
    },
    {
      "epoch": 0.9806098176313703,
      "grad_norm": 0.135556488028321,
      "learning_rate": 1.9715127877212304e-07,
      "loss": 1.0187,
      "step": 13604
    },
    {
      "epoch": 0.9806819000937071,
      "grad_norm": 0.11303762392562586,
      "learning_rate": 1.9568867192886197e-07,
      "loss": 1.1442,
      "step": 13605
    },
    {
      "epoch": 0.9807539825560441,
      "grad_norm": 0.11361962162764515,
      "learning_rate": 1.9423150531344692e-07,
      "loss": 1.111,
      "step": 13606
    },
    {
      "epoch": 0.980826065018381,
      "grad_norm": 0.1087905440211055,
      "learning_rate": 1.927797790053032e-07,
      "loss": 1.0119,
      "step": 13607
    },
    {
      "epoch": 0.9808981474807179,
      "grad_norm": 0.12103272930229161,
      "learning_rate": 1.913334930835564e-07,
      "loss": 1.0297,
      "step": 13608
    },
    {
      "epoch": 0.9809702299430548,
      "grad_norm": 0.3057163919119232,
      "learning_rate": 1.898926476270546e-07,
      "loss": 1.0206,
      "step": 13609
    },
    {
      "epoch": 0.9810423124053917,
      "grad_norm": 0.10784741500598671,
      "learning_rate": 1.8845724271432386e-07,
      "loss": 0.9819,
      "step": 13610
    },
    {
      "epoch": 0.9811143948677287,
      "grad_norm": 0.13090930754994914,
      "learning_rate": 1.8702727842361267e-07,
      "loss": 1.1117,
      "step": 13611
    },
    {
      "epoch": 0.9811864773300656,
      "grad_norm": 0.19018962865227787,
      "learning_rate": 1.8560275483286982e-07,
      "loss": 1.0079,
      "step": 13612
    },
    {
      "epoch": 0.9812585597924025,
      "grad_norm": 0.1340041229025149,
      "learning_rate": 1.841836720197443e-07,
      "loss": 1.0147,
      "step": 13613
    },
    {
      "epoch": 0.9813306422547394,
      "grad_norm": 0.11587121826752533,
      "learning_rate": 1.8277003006158532e-07,
      "loss": 1.0415,
      "step": 13614
    },
    {
      "epoch": 0.9814027247170763,
      "grad_norm": 0.12662760250445995,
      "learning_rate": 1.813618290354535e-07,
      "loss": 1.028,
      "step": 13615
    },
    {
      "epoch": 0.9814748071794133,
      "grad_norm": 0.12851039542834086,
      "learning_rate": 1.7995906901809857e-07,
      "loss": 1.0055,
      "step": 13616
    },
    {
      "epoch": 0.9815468896417502,
      "grad_norm": 0.17966988158992112,
      "learning_rate": 1.7856175008599264e-07,
      "loss": 0.9987,
      "step": 13617
    },
    {
      "epoch": 0.9816189721040871,
      "grad_norm": 0.12607531926009655,
      "learning_rate": 1.7716987231529703e-07,
      "loss": 0.9653,
      "step": 13618
    },
    {
      "epoch": 0.981691054566424,
      "grad_norm": 0.24301787968930047,
      "learning_rate": 1.7578343578188438e-07,
      "loss": 1.0411,
      "step": 13619
    },
    {
      "epoch": 0.981763137028761,
      "grad_norm": 0.11940164540946879,
      "learning_rate": 1.744024405613276e-07,
      "loss": 0.93,
      "step": 13620
    },
    {
      "epoch": 0.9818352194910979,
      "grad_norm": 0.1150147869513752,
      "learning_rate": 1.7302688672889977e-07,
      "loss": 1.0308,
      "step": 13621
    },
    {
      "epoch": 0.9819073019534347,
      "grad_norm": 0.13958429043458043,
      "learning_rate": 1.716567743595965e-07,
      "loss": 0.9567,
      "step": 13622
    },
    {
      "epoch": 0.9819793844157716,
      "grad_norm": 0.12555223691272502,
      "learning_rate": 1.7029210352808023e-07,
      "loss": 1.0474,
      "step": 13623
    },
    {
      "epoch": 0.9820514668781085,
      "grad_norm": 0.23749703497910335,
      "learning_rate": 1.6893287430874704e-07,
      "loss": 1.0193,
      "step": 13624
    },
    {
      "epoch": 0.9821235493404454,
      "grad_norm": 0.1349652220115305,
      "learning_rate": 1.6757908677568214e-07,
      "loss": 1.0362,
      "step": 13625
    },
    {
      "epoch": 0.9821956318027824,
      "grad_norm": 0.12614083985612987,
      "learning_rate": 1.662307410026931e-07,
      "loss": 1.0874,
      "step": 13626
    },
    {
      "epoch": 0.9822677142651193,
      "grad_norm": 0.16575923289821637,
      "learning_rate": 1.648878370632656e-07,
      "loss": 1.0735,
      "step": 13627
    },
    {
      "epoch": 0.9823397967274562,
      "grad_norm": 0.13600989695689716,
      "learning_rate": 1.6355037503059667e-07,
      "loss": 1.0666,
      "step": 13628
    },
    {
      "epoch": 0.9824118791897931,
      "grad_norm": 0.12338332745187332,
      "learning_rate": 1.6221835497759462e-07,
      "loss": 1.049,
      "step": 13629
    },
    {
      "epoch": 0.98248396165213,
      "grad_norm": 0.11245986317919132,
      "learning_rate": 1.6089177697686808e-07,
      "loss": 1.0752,
      "step": 13630
    },
    {
      "epoch": 0.982556044114467,
      "grad_norm": 0.17115581110511446,
      "learning_rate": 1.5957064110072584e-07,
      "loss": 1.0249,
      "step": 13631
    },
    {
      "epoch": 0.9826281265768039,
      "grad_norm": 0.12632188024672142,
      "learning_rate": 1.5825494742117697e-07,
      "loss": 0.9672,
      "step": 13632
    },
    {
      "epoch": 0.9827002090391408,
      "grad_norm": 0.12313502765427518,
      "learning_rate": 1.56944696009953e-07,
      "loss": 1.0323,
      "step": 13633
    },
    {
      "epoch": 0.9827722915014777,
      "grad_norm": 0.12924038863496112,
      "learning_rate": 1.5563988693845233e-07,
      "loss": 0.9761,
      "step": 13634
    },
    {
      "epoch": 0.9828443739638146,
      "grad_norm": 0.14002911586005212,
      "learning_rate": 1.543405202778181e-07,
      "loss": 0.9538,
      "step": 13635
    },
    {
      "epoch": 0.9829164564261516,
      "grad_norm": 0.1284710547968316,
      "learning_rate": 1.530465960988714e-07,
      "loss": 1.1163,
      "step": 13636
    },
    {
      "epoch": 0.9829885388884885,
      "grad_norm": 0.11405108087601208,
      "learning_rate": 1.517581144721447e-07,
      "loss": 1.0669,
      "step": 13637
    },
    {
      "epoch": 0.9830606213508254,
      "grad_norm": 0.17296365345311218,
      "learning_rate": 1.5047507546785966e-07,
      "loss": 1.0431,
      "step": 13638
    },
    {
      "epoch": 0.9831327038131623,
      "grad_norm": 0.11644865049493244,
      "learning_rate": 1.4919747915597138e-07,
      "loss": 1.0387,
      "step": 13639
    },
    {
      "epoch": 0.9832047862754991,
      "grad_norm": 0.14463182795633894,
      "learning_rate": 1.4792532560610195e-07,
      "loss": 1.0115,
      "step": 13640
    },
    {
      "epoch": 0.983276868737836,
      "grad_norm": 0.1145555339007759,
      "learning_rate": 1.4665861488761813e-07,
      "loss": 1.0138,
      "step": 13641
    },
    {
      "epoch": 0.983348951200173,
      "grad_norm": 0.12878292947062014,
      "learning_rate": 1.4539734706954244e-07,
      "loss": 1.0752,
      "step": 13642
    },
    {
      "epoch": 0.9834210336625099,
      "grad_norm": 0.10693695233801422,
      "learning_rate": 1.4414152222063105e-07,
      "loss": 0.944,
      "step": 13643
    },
    {
      "epoch": 0.9834931161248468,
      "grad_norm": 0.1134642113581714,
      "learning_rate": 1.4289114040936246e-07,
      "loss": 0.9953,
      "step": 13644
    },
    {
      "epoch": 0.9835651985871837,
      "grad_norm": 0.1263467871528315,
      "learning_rate": 1.4164620170386e-07,
      "loss": 1.0941,
      "step": 13645
    },
    {
      "epoch": 0.9836372810495206,
      "grad_norm": 0.12315520576844569,
      "learning_rate": 1.404067061719916e-07,
      "loss": 1.0238,
      "step": 13646
    },
    {
      "epoch": 0.9837093635118576,
      "grad_norm": 0.10851724739230022,
      "learning_rate": 1.391726538813365e-07,
      "loss": 0.9849,
      "step": 13647
    },
    {
      "epoch": 0.9837814459741945,
      "grad_norm": 0.12058889702088134,
      "learning_rate": 1.3794404489915203e-07,
      "loss": 1.1105,
      "step": 13648
    },
    {
      "epoch": 0.9838535284365314,
      "grad_norm": 0.11876614965138892,
      "learning_rate": 1.3672087929240685e-07,
      "loss": 1.0075,
      "step": 13649
    },
    {
      "epoch": 0.9839256108988683,
      "grad_norm": 0.12191012583289611,
      "learning_rate": 1.3550315712778093e-07,
      "loss": 0.9874,
      "step": 13650
    },
    {
      "epoch": 0.9839976933612052,
      "grad_norm": 0.11637858296458448,
      "learning_rate": 1.3429087847163234e-07,
      "loss": 1.1016,
      "step": 13651
    },
    {
      "epoch": 0.9840697758235422,
      "grad_norm": 0.14415258861053212,
      "learning_rate": 1.3308404339007484e-07,
      "loss": 1.0484,
      "step": 13652
    },
    {
      "epoch": 0.9841418582858791,
      "grad_norm": 0.12518664784854558,
      "learning_rate": 1.3188265194885586e-07,
      "loss": 1.0405,
      "step": 13653
    },
    {
      "epoch": 0.984213940748216,
      "grad_norm": 0.1316660961172175,
      "learning_rate": 1.3068670421348962e-07,
      "loss": 1.042,
      "step": 13654
    },
    {
      "epoch": 0.9842860232105529,
      "grad_norm": 0.12400354890886296,
      "learning_rate": 1.294962002491462e-07,
      "loss": 1.0259,
      "step": 13655
    },
    {
      "epoch": 0.9843581056728898,
      "grad_norm": 0.13216561344685124,
      "learning_rate": 1.2831114012072932e-07,
      "loss": 0.9504,
      "step": 13656
    },
    {
      "epoch": 0.9844301881352266,
      "grad_norm": 0.11151176995816099,
      "learning_rate": 1.2713152389283167e-07,
      "loss": 1.0471,
      "step": 13657
    },
    {
      "epoch": 0.9845022705975636,
      "grad_norm": 0.149149734311376,
      "learning_rate": 1.2595735162975743e-07,
      "loss": 1.1615,
      "step": 13658
    },
    {
      "epoch": 0.9845743530599005,
      "grad_norm": 0.13921260158411858,
      "learning_rate": 1.247886233954998e-07,
      "loss": 0.9999,
      "step": 13659
    },
    {
      "epoch": 0.9846464355222374,
      "grad_norm": 0.12140622413834756,
      "learning_rate": 1.2362533925377452e-07,
      "loss": 0.9931,
      "step": 13660
    },
    {
      "epoch": 0.9847185179845743,
      "grad_norm": 0.13930584155881437,
      "learning_rate": 1.224674992679975e-07,
      "loss": 1.054,
      "step": 13661
    },
    {
      "epoch": 0.9847906004469112,
      "grad_norm": 0.9828460179648515,
      "learning_rate": 1.2131510350126274e-07,
      "loss": 1.0215,
      "step": 13662
    },
    {
      "epoch": 0.9848626829092482,
      "grad_norm": 0.17016431040054542,
      "learning_rate": 1.2016815201639775e-07,
      "loss": 1.0072,
      "step": 13663
    },
    {
      "epoch": 0.9849347653715851,
      "grad_norm": 0.10588211248368178,
      "learning_rate": 1.1902664487590809e-07,
      "loss": 1.1206,
      "step": 13664
    },
    {
      "epoch": 0.985006847833922,
      "grad_norm": 0.13930473301875915,
      "learning_rate": 1.1789058214204396e-07,
      "loss": 1.1291,
      "step": 13665
    },
    {
      "epoch": 0.9850789302962589,
      "grad_norm": 0.13012556928069108,
      "learning_rate": 1.167599638767003e-07,
      "loss": 0.9618,
      "step": 13666
    },
    {
      "epoch": 0.9851510127585958,
      "grad_norm": 0.11399466638819127,
      "learning_rate": 1.1563479014152778e-07,
      "loss": 1.0547,
      "step": 13667
    },
    {
      "epoch": 0.9852230952209328,
      "grad_norm": 0.12536210045254761,
      "learning_rate": 1.1451506099784404e-07,
      "loss": 0.9898,
      "step": 13668
    },
    {
      "epoch": 0.9852951776832697,
      "grad_norm": 0.11094436831089247,
      "learning_rate": 1.1340077650670022e-07,
      "loss": 1.0245,
      "step": 13669
    },
    {
      "epoch": 0.9853672601456066,
      "grad_norm": 0.12997822459441563,
      "learning_rate": 1.1229193672881444e-07,
      "loss": 1.0806,
      "step": 13670
    },
    {
      "epoch": 0.9854393426079435,
      "grad_norm": 0.11846371078276954,
      "learning_rate": 1.1118854172464942e-07,
      "loss": 1.1184,
      "step": 13671
    },
    {
      "epoch": 0.9855114250702804,
      "grad_norm": 0.14910758255624976,
      "learning_rate": 1.1009059155433488e-07,
      "loss": 0.9946,
      "step": 13672
    },
    {
      "epoch": 0.9855835075326174,
      "grad_norm": 0.1161528434489187,
      "learning_rate": 1.089980862777229e-07,
      "loss": 1.0717,
      "step": 13673
    },
    {
      "epoch": 0.9856555899949542,
      "grad_norm": 0.1205713598350248,
      "learning_rate": 1.0791102595436587e-07,
      "loss": 1.042,
      "step": 13674
    },
    {
      "epoch": 0.9857276724572911,
      "grad_norm": 0.1450516462368885,
      "learning_rate": 1.0682941064351637e-07,
      "loss": 1.0318,
      "step": 13675
    },
    {
      "epoch": 0.985799754919628,
      "grad_norm": 0.13024838415824777,
      "learning_rate": 1.0575324040413836e-07,
      "loss": 0.9984,
      "step": 13676
    },
    {
      "epoch": 0.9858718373819649,
      "grad_norm": 0.1435150925571919,
      "learning_rate": 1.0468251529488492e-07,
      "loss": 0.9373,
      "step": 13677
    },
    {
      "epoch": 0.9859439198443019,
      "grad_norm": 0.12939998165313762,
      "learning_rate": 1.0361723537412049e-07,
      "loss": 1.0009,
      "step": 13678
    },
    {
      "epoch": 0.9860160023066388,
      "grad_norm": 0.12233736341851768,
      "learning_rate": 1.0255740069992081e-07,
      "loss": 1.0199,
      "step": 13679
    },
    {
      "epoch": 0.9860880847689757,
      "grad_norm": 0.1625882770483042,
      "learning_rate": 1.015030113300508e-07,
      "loss": 1.0009,
      "step": 13680
    },
    {
      "epoch": 0.9861601672313126,
      "grad_norm": 0.11417260395316396,
      "learning_rate": 1.0045406732197559e-07,
      "loss": 1.0105,
      "step": 13681
    },
    {
      "epoch": 0.9862322496936495,
      "grad_norm": 0.1232205341636116,
      "learning_rate": 9.941056873289389e-08,
      "loss": 1.0271,
      "step": 13682
    },
    {
      "epoch": 0.9863043321559865,
      "grad_norm": 0.17798426616574503,
      "learning_rate": 9.83725156196602e-08,
      "loss": 1.0055,
      "step": 13683
    },
    {
      "epoch": 0.9863764146183234,
      "grad_norm": 0.12113623923850356,
      "learning_rate": 9.733990803887371e-08,
      "loss": 0.9531,
      "step": 13684
    },
    {
      "epoch": 0.9864484970806603,
      "grad_norm": 0.11829482280931918,
      "learning_rate": 9.631274604682273e-08,
      "loss": 0.9913,
      "step": 13685
    },
    {
      "epoch": 0.9865205795429972,
      "grad_norm": 0.12434293601530184,
      "learning_rate": 9.529102969948467e-08,
      "loss": 1.0245,
      "step": 13686
    },
    {
      "epoch": 0.9865926620053341,
      "grad_norm": 0.14608536507431916,
      "learning_rate": 9.427475905255944e-08,
      "loss": 1.0597,
      "step": 13687
    },
    {
      "epoch": 0.986664744467671,
      "grad_norm": 0.16158815896811182,
      "learning_rate": 9.326393416144719e-08,
      "loss": 1.0382,
      "step": 13688
    },
    {
      "epoch": 0.986736826930008,
      "grad_norm": 0.1223194471887432,
      "learning_rate": 9.225855508123715e-08,
      "loss": 1.0543,
      "step": 13689
    },
    {
      "epoch": 0.9868089093923449,
      "grad_norm": 0.1332677596109808,
      "learning_rate": 9.125862186674106e-08,
      "loss": 0.9796,
      "step": 13690
    },
    {
      "epoch": 0.9868809918546818,
      "grad_norm": 0.11778225485312893,
      "learning_rate": 9.026413457244864e-08,
      "loss": 1.0365,
      "step": 13691
    },
    {
      "epoch": 0.9869530743170186,
      "grad_norm": 0.11852071888140261,
      "learning_rate": 8.92750932525832e-08,
      "loss": 0.9972,
      "step": 13692
    },
    {
      "epoch": 0.9870251567793555,
      "grad_norm": 0.15285759529527043,
      "learning_rate": 8.829149796105718e-08,
      "loss": 1.0726,
      "step": 13693
    },
    {
      "epoch": 0.9870972392416925,
      "grad_norm": 0.12954637409798614,
      "learning_rate": 8.731334875148322e-08,
      "loss": 1.0826,
      "step": 13694
    },
    {
      "epoch": 0.9871693217040294,
      "grad_norm": 0.26346126792499014,
      "learning_rate": 8.634064567716315e-08,
      "loss": 1.0685,
      "step": 13695
    },
    {
      "epoch": 0.9872414041663663,
      "grad_norm": 0.11445943506975861,
      "learning_rate": 8.537338879114342e-08,
      "loss": 1.0514,
      "step": 13696
    },
    {
      "epoch": 0.9873134866287032,
      "grad_norm": 0.1322310107694005,
      "learning_rate": 8.441157814612632e-08,
      "loss": 0.991,
      "step": 13697
    },
    {
      "epoch": 0.9873855690910401,
      "grad_norm": 0.11743787811211522,
      "learning_rate": 8.345521379455879e-08,
      "loss": 1.0294,
      "step": 13698
    },
    {
      "epoch": 0.9874576515533771,
      "grad_norm": 0.18907653471268482,
      "learning_rate": 8.250429578855467e-08,
      "loss": 1.1668,
      "step": 13699
    },
    {
      "epoch": 0.987529734015714,
      "grad_norm": 0.11610744683472173,
      "learning_rate": 8.155882417995032e-08,
      "loss": 1.028,
      "step": 13700
    },
    {
      "epoch": 0.9876018164780509,
      "grad_norm": 0.11410654328374804,
      "learning_rate": 8.061879902029335e-08,
      "loss": 1.0124,
      "step": 13701
    },
    {
      "epoch": 0.9876738989403878,
      "grad_norm": 0.13640396884758846,
      "learning_rate": 7.968422036080946e-08,
      "loss": 1.0197,
      "step": 13702
    },
    {
      "epoch": 0.9877459814027247,
      "grad_norm": 0.15523227478320825,
      "learning_rate": 7.875508825244682e-08,
      "loss": 1.0326,
      "step": 13703
    },
    {
      "epoch": 0.9878180638650617,
      "grad_norm": 0.12776951388982138,
      "learning_rate": 7.783140274585377e-08,
      "loss": 0.9699,
      "step": 13704
    },
    {
      "epoch": 0.9878901463273986,
      "grad_norm": 0.14342039421287334,
      "learning_rate": 7.691316389139003e-08,
      "loss": 1.146,
      "step": 13705
    },
    {
      "epoch": 0.9879622287897355,
      "grad_norm": 0.11924968785168112,
      "learning_rate": 7.600037173909336e-08,
      "loss": 0.9806,
      "step": 13706
    },
    {
      "epoch": 0.9880343112520724,
      "grad_norm": 0.18538665012862582,
      "learning_rate": 7.509302633871284e-08,
      "loss": 0.9787,
      "step": 13707
    },
    {
      "epoch": 0.9881063937144093,
      "grad_norm": 0.11637355587095445,
      "learning_rate": 7.419112773972003e-08,
      "loss": 0.9892,
      "step": 13708
    },
    {
      "epoch": 0.9881784761767461,
      "grad_norm": 0.12683402246743639,
      "learning_rate": 7.32946759912756e-08,
      "loss": 0.9708,
      "step": 13709
    },
    {
      "epoch": 0.9882505586390831,
      "grad_norm": 0.1436069972519469,
      "learning_rate": 7.240367114224044e-08,
      "loss": 1.0103,
      "step": 13710
    },
    {
      "epoch": 0.98832264110142,
      "grad_norm": 0.11570355987541923,
      "learning_rate": 7.151811324117575e-08,
      "loss": 0.9387,
      "step": 13711
    },
    {
      "epoch": 0.9883947235637569,
      "grad_norm": 0.12918532823936874,
      "learning_rate": 7.063800233636509e-08,
      "loss": 1.0617,
      "step": 13712
    },
    {
      "epoch": 0.9884668060260938,
      "grad_norm": 0.11037133697631996,
      "learning_rate": 6.976333847578121e-08,
      "loss": 1.0095,
      "step": 13713
    },
    {
      "epoch": 0.9885388884884307,
      "grad_norm": 0.11945935665582148,
      "learning_rate": 6.889412170708598e-08,
      "loss": 1.1012,
      "step": 13714
    },
    {
      "epoch": 0.9886109709507677,
      "grad_norm": 0.1149271135887696,
      "learning_rate": 6.803035207767483e-08,
      "loss": 0.993,
      "step": 13715
    },
    {
      "epoch": 0.9886830534131046,
      "grad_norm": 0.19882312002935884,
      "learning_rate": 6.717202963463232e-08,
      "loss": 1.0561,
      "step": 13716
    },
    {
      "epoch": 0.9887551358754415,
      "grad_norm": 0.11894066323489301,
      "learning_rate": 6.631915442473213e-08,
      "loss": 1.0823,
      "step": 13717
    },
    {
      "epoch": 0.9888272183377784,
      "grad_norm": 0.12826703838043332,
      "learning_rate": 6.547172649448152e-08,
      "loss": 1.066,
      "step": 13718
    },
    {
      "epoch": 0.9888993008001153,
      "grad_norm": 0.11708448113334391,
      "learning_rate": 6.462974589004355e-08,
      "loss": 0.9949,
      "step": 13719
    },
    {
      "epoch": 0.9889713832624523,
      "grad_norm": 0.11830808675695106,
      "learning_rate": 6.379321265734817e-08,
      "loss": 1.1009,
      "step": 13720
    },
    {
      "epoch": 0.9890434657247892,
      "grad_norm": 0.12773936795395036,
      "learning_rate": 6.296212684197e-08,
      "loss": 1.0947,
      "step": 13721
    },
    {
      "epoch": 0.9891155481871261,
      "grad_norm": 0.1162879197562804,
      "learning_rate": 6.213648848921728e-08,
      "loss": 1.0603,
      "step": 13722
    },
    {
      "epoch": 0.989187630649463,
      "grad_norm": 0.17835461203988862,
      "learning_rate": 6.131629764409841e-08,
      "loss": 1.1056,
      "step": 13723
    },
    {
      "epoch": 0.9892597131117999,
      "grad_norm": 0.12189026250307045,
      "learning_rate": 6.05015543513221e-08,
      "loss": 1.0227,
      "step": 13724
    },
    {
      "epoch": 0.9893317955741369,
      "grad_norm": 0.29466657765064236,
      "learning_rate": 5.969225865528615e-08,
      "loss": 1.0329,
      "step": 13725
    },
    {
      "epoch": 0.9894038780364737,
      "grad_norm": 0.11258382149285125,
      "learning_rate": 5.888841060012196e-08,
      "loss": 1.0257,
      "step": 13726
    },
    {
      "epoch": 0.9894759604988106,
      "grad_norm": 0.1283304983413966,
      "learning_rate": 5.809001022962779e-08,
      "loss": 1.0451,
      "step": 13727
    },
    {
      "epoch": 0.9895480429611475,
      "grad_norm": 0.11145472377147085,
      "learning_rate": 5.7297057587346604e-08,
      "loss": 0.9919,
      "step": 13728
    },
    {
      "epoch": 0.9896201254234844,
      "grad_norm": 0.12663157122194246,
      "learning_rate": 5.6509552716477176e-08,
      "loss": 1.0811,
      "step": 13729
    },
    {
      "epoch": 0.9896922078858214,
      "grad_norm": 0.11928720842073973,
      "learning_rate": 5.572749565996294e-08,
      "loss": 1.0232,
      "step": 13730
    },
    {
      "epoch": 0.9897642903481583,
      "grad_norm": 0.14730793122881702,
      "learning_rate": 5.495088646042534e-08,
      "loss": 0.9518,
      "step": 13731
    },
    {
      "epoch": 0.9898363728104952,
      "grad_norm": 0.12135552094721103,
      "learning_rate": 5.41797251602083e-08,
      "loss": 1.0424,
      "step": 13732
    },
    {
      "epoch": 0.9899084552728321,
      "grad_norm": 0.1134774148187126,
      "learning_rate": 5.341401180132266e-08,
      "loss": 1.0755,
      "step": 13733
    },
    {
      "epoch": 0.989980537735169,
      "grad_norm": 0.134855878941979,
      "learning_rate": 5.2653746425535e-08,
      "loss": 1.0609,
      "step": 13734
    },
    {
      "epoch": 0.990052620197506,
      "grad_norm": 0.11393107721104855,
      "learning_rate": 5.189892907425664e-08,
      "loss": 1.031,
      "step": 13735
    },
    {
      "epoch": 0.9901247026598429,
      "grad_norm": 0.149943493054864,
      "learning_rate": 5.1149559788654655e-08,
      "loss": 0.9902,
      "step": 13736
    },
    {
      "epoch": 0.9901967851221798,
      "grad_norm": 0.14251083481796917,
      "learning_rate": 5.040563860957415e-08,
      "loss": 1.1348,
      "step": 13737
    },
    {
      "epoch": 0.9902688675845167,
      "grad_norm": 0.1651824891835939,
      "learning_rate": 4.966716557756046e-08,
      "loss": 0.9994,
      "step": 13738
    },
    {
      "epoch": 0.9903409500468536,
      "grad_norm": 0.12504143705625975,
      "learning_rate": 4.893414073285918e-08,
      "loss": 1.0976,
      "step": 13739
    },
    {
      "epoch": 0.9904130325091905,
      "grad_norm": 0.10885557884963633,
      "learning_rate": 4.820656411543834e-08,
      "loss": 1.0093,
      "step": 13740
    },
    {
      "epoch": 0.9904851149715275,
      "grad_norm": 0.11350261758854265,
      "learning_rate": 4.7484435764966195e-08,
      "loss": 1.0644,
      "step": 13741
    },
    {
      "epoch": 0.9905571974338644,
      "grad_norm": 0.15327982762619902,
      "learning_rate": 4.676775572077796e-08,
      "loss": 1.0358,
      "step": 13742
    },
    {
      "epoch": 0.9906292798962013,
      "grad_norm": 1.1058984953356361,
      "learning_rate": 4.6056524021964584e-08,
      "loss": 1.017,
      "step": 13743
    },
    {
      "epoch": 0.9907013623585381,
      "grad_norm": 0.17392761130263934,
      "learning_rate": 4.535074070727285e-08,
      "loss": 0.9634,
      "step": 13744
    },
    {
      "epoch": 0.990773444820875,
      "grad_norm": 0.10427591923433974,
      "learning_rate": 4.4650405815194196e-08,
      "loss": 1.0306,
      "step": 13745
    },
    {
      "epoch": 0.990845527283212,
      "grad_norm": 0.12631471166832728,
      "learning_rate": 4.395551938389808e-08,
      "loss": 1.0558,
      "step": 13746
    },
    {
      "epoch": 0.9909176097455489,
      "grad_norm": 0.13408919895800056,
      "learning_rate": 4.326608145124311e-08,
      "loss": 1.0648,
      "step": 13747
    },
    {
      "epoch": 0.9909896922078858,
      "grad_norm": 0.13217691613059582,
      "learning_rate": 4.258209205484365e-08,
      "loss": 1.0654,
      "step": 13748
    },
    {
      "epoch": 0.9910617746702227,
      "grad_norm": 0.13494564092517475,
      "learning_rate": 4.1903551231947666e-08,
      "loss": 1.0842,
      "step": 13749
    },
    {
      "epoch": 0.9911338571325596,
      "grad_norm": 0.1599099365568229,
      "learning_rate": 4.123045901957001e-08,
      "loss": 0.9715,
      "step": 13750
    },
    {
      "epoch": 0.9912059395948966,
      "grad_norm": 0.11653493542321254,
      "learning_rate": 4.056281545439245e-08,
      "loss": 1.0774,
      "step": 13751
    },
    {
      "epoch": 0.9912780220572335,
      "grad_norm": 0.15260126881852942,
      "learning_rate": 3.990062057278587e-08,
      "loss": 1.0745,
      "step": 13752
    },
    {
      "epoch": 0.9913501045195704,
      "grad_norm": 0.12434631455012239,
      "learning_rate": 3.924387441087696e-08,
      "loss": 1.0016,
      "step": 13753
    },
    {
      "epoch": 0.9914221869819073,
      "grad_norm": 0.12263951643049885,
      "learning_rate": 3.8592577004459285e-08,
      "loss": 1.0346,
      "step": 13754
    },
    {
      "epoch": 0.9914942694442442,
      "grad_norm": 0.11389086276085965,
      "learning_rate": 3.7946728389015586e-08,
      "loss": 1.06,
      "step": 13755
    },
    {
      "epoch": 0.9915663519065812,
      "grad_norm": 0.12861623857388443,
      "learning_rate": 3.730632859976213e-08,
      "loss": 1.0577,
      "step": 13756
    },
    {
      "epoch": 0.9916384343689181,
      "grad_norm": 0.11785065814284187,
      "learning_rate": 3.667137767160433e-08,
      "loss": 1.038,
      "step": 13757
    },
    {
      "epoch": 0.991710516831255,
      "grad_norm": 0.1242511440591323,
      "learning_rate": 3.604187563915895e-08,
      "loss": 0.9783,
      "step": 13758
    },
    {
      "epoch": 0.9917825992935919,
      "grad_norm": 0.132219566472861,
      "learning_rate": 3.541782253673187e-08,
      "loss": 0.9964,
      "step": 13759
    },
    {
      "epoch": 0.9918546817559288,
      "grad_norm": 0.24424901074328823,
      "learning_rate": 3.479921839834033e-08,
      "loss": 0.9917,
      "step": 13760
    },
    {
      "epoch": 0.9919267642182656,
      "grad_norm": 0.11530652619548303,
      "learning_rate": 3.4186063257701796e-08,
      "loss": 1.034,
      "step": 13761
    },
    {
      "epoch": 0.9919988466806026,
      "grad_norm": 0.13655222125279187,
      "learning_rate": 3.357835714825619e-08,
      "loss": 1.0661,
      "step": 13762
    },
    {
      "epoch": 0.9920709291429395,
      "grad_norm": 0.1313519248316184,
      "learning_rate": 3.297610010311036e-08,
      "loss": 1.0155,
      "step": 13763
    },
    {
      "epoch": 0.9921430116052764,
      "grad_norm": 0.1211914620450294,
      "learning_rate": 3.237929215509361e-08,
      "loss": 1.0296,
      "step": 13764
    },
    {
      "epoch": 0.9922150940676133,
      "grad_norm": 0.11155365567804948,
      "learning_rate": 3.178793333673546e-08,
      "loss": 1.0533,
      "step": 13765
    },
    {
      "epoch": 0.9922871765299502,
      "grad_norm": 0.12196150361575264,
      "learning_rate": 3.120202368028791e-08,
      "loss": 0.9246,
      "step": 13766
    },
    {
      "epoch": 0.9923592589922872,
      "grad_norm": 0.20363715697245846,
      "learning_rate": 3.062156321766985e-08,
      "loss": 1.0503,
      "step": 13767
    },
    {
      "epoch": 0.9924313414546241,
      "grad_norm": 0.12213111353890259,
      "learning_rate": 3.004655198053374e-08,
      "loss": 0.9792,
      "step": 13768
    },
    {
      "epoch": 0.992503423916961,
      "grad_norm": 0.1475529291924483,
      "learning_rate": 2.947699000021009e-08,
      "loss": 1.0454,
      "step": 13769
    },
    {
      "epoch": 0.9925755063792979,
      "grad_norm": 0.11587919599450329,
      "learning_rate": 2.8912877307751828e-08,
      "loss": 1.0547,
      "step": 13770
    },
    {
      "epoch": 0.9926475888416348,
      "grad_norm": 0.12925514278833197,
      "learning_rate": 2.8354213933912133e-08,
      "loss": 1.0453,
      "step": 13771
    },
    {
      "epoch": 0.9927196713039718,
      "grad_norm": 0.1267803686834243,
      "learning_rate": 2.780099990914442e-08,
      "loss": 1.0257,
      "step": 13772
    },
    {
      "epoch": 0.9927917537663087,
      "grad_norm": 0.13187784326157176,
      "learning_rate": 2.7253235263591248e-08,
      "loss": 1.0447,
      "step": 13773
    },
    {
      "epoch": 0.9928638362286456,
      "grad_norm": 0.11956675151051607,
      "learning_rate": 2.6710920027117615e-08,
      "loss": 1.0874,
      "step": 13774
    },
    {
      "epoch": 0.9929359186909825,
      "grad_norm": 0.12925486150650403,
      "learning_rate": 2.617405422929986e-08,
      "loss": 1.0896,
      "step": 13775
    },
    {
      "epoch": 0.9930080011533194,
      "grad_norm": 0.11582500775815355,
      "learning_rate": 2.564263789937016e-08,
      "loss": 1.0222,
      "step": 13776
    },
    {
      "epoch": 0.9930800836156564,
      "grad_norm": 0.12406413523103423,
      "learning_rate": 2.5116671066316433e-08,
      "loss": 1.0867,
      "step": 13777
    },
    {
      "epoch": 0.9931521660779932,
      "grad_norm": 0.11524664146330006,
      "learning_rate": 2.4596153758815743e-08,
      "loss": 1.018,
      "step": 13778
    },
    {
      "epoch": 0.9932242485403301,
      "grad_norm": 0.1243364675398403,
      "learning_rate": 2.4081086005223186e-08,
      "loss": 1.0142,
      "step": 13779
    },
    {
      "epoch": 0.993296331002667,
      "grad_norm": 0.13019104394651002,
      "learning_rate": 2.3571467833616302e-08,
      "loss": 1.0349,
      "step": 13780
    },
    {
      "epoch": 0.9933684134650039,
      "grad_norm": 0.1346919494740404,
      "learning_rate": 2.306729927178397e-08,
      "loss": 1.0463,
      "step": 13781
    },
    {
      "epoch": 0.9934404959273408,
      "grad_norm": 0.18206618715668896,
      "learning_rate": 2.2568580347204215e-08,
      "loss": 1.0306,
      "step": 13782
    },
    {
      "epoch": 0.9935125783896778,
      "grad_norm": 0.111574785448388,
      "learning_rate": 2.2075311087066398e-08,
      "loss": 1.0956,
      "step": 13783
    },
    {
      "epoch": 0.9935846608520147,
      "grad_norm": 0.13014760144594512,
      "learning_rate": 2.1587491518237913e-08,
      "loss": 1.0532,
      "step": 13784
    },
    {
      "epoch": 0.9936567433143516,
      "grad_norm": 0.118609029332079,
      "learning_rate": 2.1105121667330806e-08,
      "loss": 1.0696,
      "step": 13785
    },
    {
      "epoch": 0.9937288257766885,
      "grad_norm": 0.13261711827403966,
      "learning_rate": 2.0628201560635162e-08,
      "loss": 1.119,
      "step": 13786
    },
    {
      "epoch": 0.9938009082390254,
      "grad_norm": 0.12704597325541694,
      "learning_rate": 2.01567312241413e-08,
      "loss": 1.0379,
      "step": 13787
    },
    {
      "epoch": 0.9938729907013624,
      "grad_norm": 0.14277794658154935,
      "learning_rate": 1.9690710683550884e-08,
      "loss": 1.0142,
      "step": 13788
    },
    {
      "epoch": 0.9939450731636993,
      "grad_norm": 0.125466897654732,
      "learning_rate": 1.9230139964265813e-08,
      "loss": 1.0268,
      "step": 13789
    },
    {
      "epoch": 0.9940171556260362,
      "grad_norm": 0.11824293145015285,
      "learning_rate": 1.8775019091388235e-08,
      "loss": 1.0879,
      "step": 13790
    },
    {
      "epoch": 0.9940892380883731,
      "grad_norm": 0.1224855232611173,
      "learning_rate": 1.8325348089731632e-08,
      "loss": 1.025,
      "step": 13791
    },
    {
      "epoch": 0.99416132055071,
      "grad_norm": 0.12085328381697877,
      "learning_rate": 1.7881126983809727e-08,
      "loss": 1.0824,
      "step": 13792
    },
    {
      "epoch": 0.994233403013047,
      "grad_norm": 0.13608377834205362,
      "learning_rate": 1.744235579782538e-08,
      "loss": 1.0314,
      "step": 13793
    },
    {
      "epoch": 0.9943054854753839,
      "grad_norm": 0.1312463117651486,
      "learning_rate": 1.7009034555703907e-08,
      "loss": 1.0077,
      "step": 13794
    },
    {
      "epoch": 0.9943775679377208,
      "grad_norm": 0.11546269770571944,
      "learning_rate": 1.6581163281059743e-08,
      "loss": 0.9965,
      "step": 13795
    },
    {
      "epoch": 0.9944496504000576,
      "grad_norm": 0.1177034848096324,
      "learning_rate": 1.6158741997229775e-08,
      "loss": 1.0833,
      "step": 13796
    },
    {
      "epoch": 0.9945217328623945,
      "grad_norm": 0.1376025549079062,
      "learning_rate": 1.574177072721783e-08,
      "loss": 0.9603,
      "step": 13797
    },
    {
      "epoch": 0.9945938153247315,
      "grad_norm": 0.12637423160117894,
      "learning_rate": 1.5330249493772376e-08,
      "loss": 1.0406,
      "step": 13798
    },
    {
      "epoch": 0.9946658977870684,
      "grad_norm": 0.10655193501580563,
      "learning_rate": 1.4924178319319916e-08,
      "loss": 0.9825,
      "step": 13799
    },
    {
      "epoch": 0.9947379802494053,
      "grad_norm": 0.11761758892737947,
      "learning_rate": 1.4523557225976093e-08,
      "loss": 0.9248,
      "step": 13800
    },
    {
      "epoch": 0.9948100627117422,
      "grad_norm": 0.1422293400604382,
      "learning_rate": 1.41283862356123e-08,
      "loss": 1.0606,
      "step": 13801
    },
    {
      "epoch": 0.9948821451740791,
      "grad_norm": 0.12416209752655873,
      "learning_rate": 1.3738665369733561e-08,
      "loss": 1.0703,
      "step": 13802
    },
    {
      "epoch": 0.994954227636416,
      "grad_norm": 0.13163896560906588,
      "learning_rate": 1.3354394649611745e-08,
      "loss": 1.0346,
      "step": 13803
    },
    {
      "epoch": 0.995026310098753,
      "grad_norm": 0.11802183583342851,
      "learning_rate": 1.2975574096185661e-08,
      "loss": 1.0292,
      "step": 13804
    },
    {
      "epoch": 0.9950983925610899,
      "grad_norm": 0.14619270956874253,
      "learning_rate": 1.260220373008325e-08,
      "loss": 1.01,
      "step": 13805
    },
    {
      "epoch": 0.9951704750234268,
      "grad_norm": 0.1326434446418669,
      "learning_rate": 1.223428357168821e-08,
      "loss": 1.0605,
      "step": 13806
    },
    {
      "epoch": 0.9952425574857637,
      "grad_norm": 0.12047926365448042,
      "learning_rate": 1.1871813641040064e-08,
      "loss": 1.0868,
      "step": 13807
    },
    {
      "epoch": 0.9953146399481007,
      "grad_norm": 0.12807003500725211,
      "learning_rate": 1.151479395788968e-08,
      "loss": 1.0157,
      "step": 13808
    },
    {
      "epoch": 0.9953867224104376,
      "grad_norm": 0.1181994148260149,
      "learning_rate": 1.1163224541710371e-08,
      "loss": 1.0005,
      "step": 13809
    },
    {
      "epoch": 0.9954588048727745,
      "grad_norm": 0.1268927541566811,
      "learning_rate": 1.0817105411664586e-08,
      "loss": 1.004,
      "step": 13810
    },
    {
      "epoch": 0.9955308873351114,
      "grad_norm": 0.1459313547019061,
      "learning_rate": 1.047643658660391e-08,
      "loss": 1.0788,
      "step": 13811
    },
    {
      "epoch": 0.9956029697974483,
      "grad_norm": 0.12255733604159058,
      "learning_rate": 1.014121808511348e-08,
      "loss": 0.9179,
      "step": 13812
    },
    {
      "epoch": 0.9956750522597851,
      "grad_norm": 0.1255809433214667,
      "learning_rate": 9.811449925467565e-09,
      "loss": 0.9898,
      "step": 13813
    },
    {
      "epoch": 0.9957471347221221,
      "grad_norm": 0.12377883045880014,
      "learning_rate": 9.487132125629572e-09,
      "loss": 1.0329,
      "step": 13814
    },
    {
      "epoch": 0.995819217184459,
      "grad_norm": 0.1249122925693833,
      "learning_rate": 9.168264703285356e-09,
      "loss": 1.0928,
      "step": 13815
    },
    {
      "epoch": 0.9958912996467959,
      "grad_norm": 0.16318318924905825,
      "learning_rate": 8.85484767582101e-09,
      "loss": 1.0618,
      "step": 13816
    },
    {
      "epoch": 0.9959633821091328,
      "grad_norm": 0.20870550385414777,
      "learning_rate": 8.546881060311762e-09,
      "loss": 1.0708,
      "step": 13817
    },
    {
      "epoch": 0.9960354645714697,
      "grad_norm": 0.18323160641270014,
      "learning_rate": 8.244364873533084e-09,
      "loss": 1.0333,
      "step": 13818
    },
    {
      "epoch": 0.9961075470338067,
      "grad_norm": 0.1437784976524841,
      "learning_rate": 7.947299132005093e-09,
      "loss": 0.9984,
      "step": 13819
    },
    {
      "epoch": 0.9961796294961436,
      "grad_norm": 0.15294745374990784,
      "learning_rate": 7.655683851892637e-09,
      "loss": 1.0254,
      "step": 13820
    },
    {
      "epoch": 0.9962517119584805,
      "grad_norm": 0.1404429527768675,
      "learning_rate": 7.369519049105211e-09,
      "loss": 0.9937,
      "step": 13821
    },
    {
      "epoch": 0.9963237944208174,
      "grad_norm": 0.11745553684962494,
      "learning_rate": 7.08880473924145e-09,
      "loss": 1.0592,
      "step": 13822
    },
    {
      "epoch": 0.9963958768831543,
      "grad_norm": 0.11580782128293457,
      "learning_rate": 6.813540937600227e-09,
      "loss": 1.0362,
      "step": 13823
    },
    {
      "epoch": 0.9964679593454913,
      "grad_norm": 0.11667246258491026,
      "learning_rate": 6.543727659180654e-09,
      "loss": 1.0145,
      "step": 13824
    },
    {
      "epoch": 0.9965400418078282,
      "grad_norm": 0.14023851909205448,
      "learning_rate": 6.279364918704289e-09,
      "loss": 1.0641,
      "step": 13825
    },
    {
      "epoch": 0.9966121242701651,
      "grad_norm": 0.11874245560825784,
      "learning_rate": 6.020452730559623e-09,
      "loss": 1.0972,
      "step": 13826
    },
    {
      "epoch": 0.996684206732502,
      "grad_norm": 0.12060485707495605,
      "learning_rate": 5.766991108879793e-09,
      "loss": 0.9444,
      "step": 13827
    },
    {
      "epoch": 0.9967562891948389,
      "grad_norm": 0.1176704633187657,
      "learning_rate": 5.518980067475976e-09,
      "loss": 0.996,
      "step": 13828
    },
    {
      "epoch": 0.9968283716571759,
      "grad_norm": 0.11357532098917088,
      "learning_rate": 5.276419619848483e-09,
      "loss": 1.0015,
      "step": 13829
    },
    {
      "epoch": 0.9969004541195127,
      "grad_norm": 0.13727860535017652,
      "learning_rate": 5.039309779253376e-09,
      "loss": 1.09,
      "step": 13830
    },
    {
      "epoch": 0.9969725365818496,
      "grad_norm": 0.13292658005665256,
      "learning_rate": 4.8076505585803455e-09,
      "loss": 1.0856,
      "step": 13831
    },
    {
      "epoch": 0.9970446190441865,
      "grad_norm": 0.1172611336474974,
      "learning_rate": 4.581441970485934e-09,
      "loss": 1.1166,
      "step": 13832
    },
    {
      "epoch": 0.9971167015065234,
      "grad_norm": 0.1831127067558302,
      "learning_rate": 4.360684027282513e-09,
      "loss": 1.0023,
      "step": 13833
    },
    {
      "epoch": 0.9971887839688603,
      "grad_norm": 0.11482268148429169,
      "learning_rate": 4.145376741004902e-09,
      "loss": 0.9879,
      "step": 13834
    },
    {
      "epoch": 0.9972608664311973,
      "grad_norm": 0.12281877141305131,
      "learning_rate": 3.935520123399261e-09,
      "loss": 1.0551,
      "step": 13835
    },
    {
      "epoch": 0.9973329488935342,
      "grad_norm": 0.12379235048447179,
      "learning_rate": 3.731114185889783e-09,
      "loss": 1.0223,
      "step": 13836
    },
    {
      "epoch": 0.9974050313558711,
      "grad_norm": 0.11793949676681441,
      "learning_rate": 3.5321589396342114e-09,
      "loss": 1.0255,
      "step": 13837
    },
    {
      "epoch": 0.997477113818208,
      "grad_norm": 0.12113679272787288,
      "learning_rate": 3.3386543954683212e-09,
      "loss": 0.9919,
      "step": 13838
    },
    {
      "epoch": 0.997549196280545,
      "grad_norm": 0.12981730642117192,
      "learning_rate": 3.150600563939232e-09,
      "loss": 1.0926,
      "step": 13839
    },
    {
      "epoch": 0.9976212787428819,
      "grad_norm": 0.13677602338181458,
      "learning_rate": 2.9679974553054046e-09,
      "loss": 0.9753,
      "step": 13840
    },
    {
      "epoch": 0.9976933612052188,
      "grad_norm": 0.11391164525489714,
      "learning_rate": 2.790845079503335e-09,
      "loss": 0.9544,
      "step": 13841
    },
    {
      "epoch": 0.9977654436675557,
      "grad_norm": 0.17749226249812644,
      "learning_rate": 2.619143446214167e-09,
      "loss": 0.9708,
      "step": 13842
    },
    {
      "epoch": 0.9978375261298926,
      "grad_norm": 0.13701508683991526,
      "learning_rate": 2.4528925647748778e-09,
      "loss": 1.0852,
      "step": 13843
    },
    {
      "epoch": 0.9979096085922295,
      "grad_norm": 0.10547807802034907,
      "learning_rate": 2.292092444255989e-09,
      "loss": 0.9946,
      "step": 13844
    },
    {
      "epoch": 0.9979816910545665,
      "grad_norm": 0.20786747926449073,
      "learning_rate": 2.1367430934282618e-09,
      "loss": 1.0315,
      "step": 13845
    },
    {
      "epoch": 0.9980537735169034,
      "grad_norm": 0.1302711147298761,
      "learning_rate": 1.986844520751596e-09,
      "loss": 1.0117,
      "step": 13846
    },
    {
      "epoch": 0.9981258559792403,
      "grad_norm": 0.127725787603276,
      "learning_rate": 1.842396734408336e-09,
      "loss": 0.9685,
      "step": 13847
    },
    {
      "epoch": 0.9981979384415771,
      "grad_norm": 0.12543550824935387,
      "learning_rate": 1.7033997422477577e-09,
      "loss": 1.0034,
      "step": 13848
    },
    {
      "epoch": 0.998270020903914,
      "grad_norm": 0.12047634111764698,
      "learning_rate": 1.5698535518748892e-09,
      "loss": 1.0255,
      "step": 13849
    },
    {
      "epoch": 0.998342103366251,
      "grad_norm": 0.1280619122137206,
      "learning_rate": 1.441758170550589e-09,
      "loss": 1.0025,
      "step": 13850
    },
    {
      "epoch": 0.9984141858285879,
      "grad_norm": 0.1334055115069854,
      "learning_rate": 1.319113605269262e-09,
      "loss": 1.0732,
      "step": 13851
    },
    {
      "epoch": 0.9984862682909248,
      "grad_norm": 0.11388977153707448,
      "learning_rate": 1.201919862703349e-09,
      "loss": 1.0128,
      "step": 13852
    },
    {
      "epoch": 0.9985583507532617,
      "grad_norm": 0.11455624749633762,
      "learning_rate": 1.0901769492477344e-09,
      "loss": 1.028,
      "step": 13853
    },
    {
      "epoch": 0.9986304332155986,
      "grad_norm": 0.13999731443895067,
      "learning_rate": 9.838848709975423e-10,
      "loss": 0.9481,
      "step": 13854
    },
    {
      "epoch": 0.9987025156779356,
      "grad_norm": 0.11943321592506835,
      "learning_rate": 8.830436337370351e-10,
      "loss": 1.0494,
      "step": 13855
    },
    {
      "epoch": 0.9987745981402725,
      "grad_norm": 0.13223938498018495,
      "learning_rate": 7.876532429729189e-10,
      "loss": 1.0419,
      "step": 13856
    },
    {
      "epoch": 0.9988466806026094,
      "grad_norm": 0.19185908479945504,
      "learning_rate": 6.977137039010373e-10,
      "loss": 1.0897,
      "step": 13857
    },
    {
      "epoch": 0.9989187630649463,
      "grad_norm": 0.1158316982070863,
      "learning_rate": 6.13225021417474e-10,
      "loss": 0.9649,
      "step": 13858
    },
    {
      "epoch": 0.9989908455272832,
      "grad_norm": 0.12624249185129216,
      "learning_rate": 5.341872001407566e-10,
      "loss": 0.9759,
      "step": 13859
    },
    {
      "epoch": 0.9990629279896202,
      "grad_norm": 0.13615420422600621,
      "learning_rate": 4.606002443674484e-10,
      "loss": 1.0209,
      "step": 13860
    },
    {
      "epoch": 0.9991350104519571,
      "grad_norm": 0.13488108224366824,
      "learning_rate": 3.9246415810545446e-10,
      "loss": 1.0343,
      "step": 13861
    },
    {
      "epoch": 0.999207092914294,
      "grad_norm": 0.415556302578774,
      "learning_rate": 3.297789450851241e-10,
      "loss": 1.0172,
      "step": 13862
    },
    {
      "epoch": 0.9992791753766309,
      "grad_norm": 0.14634474964785288,
      "learning_rate": 2.72544608714842e-10,
      "loss": 1.0155,
      "step": 13863
    },
    {
      "epoch": 0.9993512578389678,
      "grad_norm": 0.12188777338944895,
      "learning_rate": 2.207611521143349e-10,
      "loss": 0.963,
      "step": 13864
    },
    {
      "epoch": 0.9994233403013046,
      "grad_norm": 0.11846595931421156,
      "learning_rate": 1.7442857810356927e-10,
      "loss": 1.0169,
      "step": 13865
    },
    {
      "epoch": 0.9994954227636416,
      "grad_norm": 0.13179090229673174,
      "learning_rate": 1.3354688921385362e-10,
      "loss": 1.0791,
      "step": 13866
    },
    {
      "epoch": 0.9995675052259785,
      "grad_norm": 0.1332338088583456,
      "learning_rate": 9.811608766563396e-11,
      "loss": 0.9726,
      "step": 13867
    },
    {
      "epoch": 0.9996395876883154,
      "grad_norm": 0.11823059506754889,
      "learning_rate": 6.813617539069839e-11,
      "loss": 0.9766,
      "step": 13868
    },
    {
      "epoch": 0.9997116701506523,
      "grad_norm": 0.12354524180946203,
      "learning_rate": 4.360715403217697e-11,
      "loss": 1.0697,
      "step": 13869
    },
    {
      "epoch": 0.9997837526129892,
      "grad_norm": 0.11308255951054225,
      "learning_rate": 2.4529024922337328e-11,
      "loss": 1.0464,
      "step": 13870
    },
    {
      "epoch": 0.9998558350753262,
      "grad_norm": 0.12690930405320802,
      "learning_rate": 1.0901789104789117e-11,
      "loss": 0.9915,
      "step": 13871
    },
    {
      "epoch": 0.9999279175376631,
      "grad_norm": 0.1316809279102382,
      "learning_rate": 2.7254473122795276e-12,
      "loss": 0.9804,
      "step": 13872
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1382346000658165,
      "learning_rate": 0.0,
      "loss": 1.0901,
      "step": 13873
    },
    {
      "epoch": 1.0,
      "step": 13873,
      "total_flos": 1.2098491289239552e+16,
      "train_loss": 1.082582399211483,
      "train_runtime": 165481.6665,
      "train_samples_per_second": 10.73,
      "train_steps_per_second": 0.084
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 13873,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 300,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.2098491289239552e+16,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}